{"repo": "agno-agi/docs", "files": {"examples/reasoning/agents/trolley-problem.mdx": {"2c5a11695c8a13de4b8cf1be164317854a9642de": {"t": 1785746500.5485957, "n": 3151, "h": "c9822f25d9a3ef9a", "c": "---\ntitle: \"Trolley Problem Analysis\"\ndescription: \"Solve the trolley problem with ASCII diagrams using a built-in chain-of-thought agent and a deepseek-reasoner-backed agent.\"\nsource: cookbook/10_reasoning/agents/trolley_problem.py\n---\n\nDemonstrates built-in and DeepSeek-backed reasoning for ethical analysis.\n\n```python trolley_problem.py\n\"\"\"\nTrolley Problem Analysis\n========================\n\nDemonstrates built-in and DeepSeek-backed reasoning for ethical analysis.\n\"\"\"\n\nfrom agno.agent import Agent\nfrom agno.models.deepseek import DeepSeek\nfrom agno.models.openai import OpenAIChat\n\n# ---------------------------------------------------------------------------\n# Create Agents\n# ---------------------------------------------------------------------------\ncot_prompt = (\n    \"Solve the trolley problem. Evaluate multiple ethical frameworks. \"\n    \"Include an ASCII diagram of your solution.\"\n)\n\ndeepseek_prompt = (\n    \"You are a philosopher tasked with analyzing the classic 'Trolley Problem'. In this scenario, a runaway trolley \"\n    \"is barreling down the tracks towards five people who are tied up and unable to move. You are standing next to \"\n    \"a large stranger on a footbridge above the tracks. The only way to save the five people is to push this stranger \"\n    \"off the bridge onto the tracks below. This will kill the stranger, but save the five people on the tracks. \"\n    \"Should you push the stranger to save the five people? Provide a well-reasoned answer considering utilitarian, \"\n    \"deontological, and virtue ethics frameworks. \"\n    \"Include a simple ASCII art diagram to illustrate the scenario.\"\n)\n\ncot_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning=True,\n    markdown=True,\n)\n\ndeepseek_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning_model=DeepSeek(id=\"deepseek-reasoner\"),\n    markdown=True,\n)\n\n# ---------------------------------------------------------------------------\n# Run Agents\n# ---------------------------------------------------------------------------\nif __name__ == \"__main__\":\n    print(\"=== Built-in Chain Of Thought ===\")\n    cot_agent.print_response(cot_prompt, stream=True, show_full_reasoning=True)\n\n    print(\"\\n=== DeepSeek Reasoning Model ===\")\n    deepseek_agent.print_response(deepseek_prompt, stream=True)\n```\n\n## Run the Example\n\n<Steps>\n  <Snippet file=\"create-venv-step.mdx\" />\n\n  <Step title=\"Install dependencies\">\n    ```bash\n    uv pip install -U agno openai\n    ```\n  </Step>\n\n  <Step title=\"Export your API keys\">\n    <CodeGroup>\n    ```bash Mac/Linux\n    export DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    export OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n\n    ```bash Windows\n    $Env:DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    $Env:OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n    </CodeGroup>\n  </Step>\n\n  <Step title=\"Run the example\">\n    Save the code above as `trolley_problem.py`, then run:\n    ```bash\n    python trolley_problem.py\n    ```\n  </Step>\n</Steps>\n\nFull source: [cookbook/10_reasoning/agents/trolley_problem.py](https://github.com/agno-agi/agno/blob/main/cookbook/10_reasoning/agents/trolley_problem.py)\n"}, "6917c89afd1df7aa2a58b9b9c63f9c8cc19a354e": {"t": 1785764992.515693, "n": 3151, "h": "c9822f25d9a3ef9a", "c": "---\ntitle: \"Trolley Problem Analysis\"\ndescription: \"Solve the trolley problem with ASCII diagrams using a built-in chain-of-thought agent and a deepseek-reasoner-backed agent.\"\nsource: cookbook/10_reasoning/agents/trolley_problem.py\n---\n\nDemonstrates built-in and DeepSeek-backed reasoning for ethical analysis.\n\n```python trolley_problem.py\n\"\"\"\nTrolley Problem Analysis\n========================\n\nDemonstrates built-in and DeepSeek-backed reasoning for ethical analysis.\n\"\"\"\n\nfrom agno.agent import Agent\nfrom agno.models.deepseek import DeepSeek\nfrom agno.models.openai import OpenAIChat\n\n# ---------------------------------------------------------------------------\n# Create Agents\n# ---------------------------------------------------------------------------\ncot_prompt = (\n    \"Solve the trolley problem. Evaluate multiple ethical frameworks. \"\n    \"Include an ASCII diagram of your solution.\"\n)\n\ndeepseek_prompt = (\n    \"You are a philosopher tasked with analyzing the classic 'Trolley Problem'. In this scenario, a runaway trolley \"\n    \"is barreling down the tracks towards five people who are tied up and unable to move. You are standing next to \"\n    \"a large stranger on a footbridge above the tracks. The only way to save the five people is to push this stranger \"\n    \"off the bridge onto the tracks below. This will kill the stranger, but save the five people on the tracks. \"\n    \"Should you push the stranger to save the five people? Provide a well-reasoned answer considering utilitarian, \"\n    \"deontological, and virtue ethics frameworks. \"\n    \"Include a simple ASCII art diagram to illustrate the scenario.\"\n)\n\ncot_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning=True,\n    markdown=True,\n)\n\ndeepseek_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning_model=DeepSeek(id=\"deepseek-reasoner\"),\n    markdown=True,\n)\n\n# ---------------------------------------------------------------------------\n# Run Agents\n# ---------------------------------------------------------------------------\nif __name__ == \"__main__\":\n    print(\"=== Built-in Chain Of Thought ===\")\n    cot_agent.print_response(cot_prompt, stream=True, show_full_reasoning=True)\n\n    print(\"\\n=== DeepSeek Reasoning Model ===\")\n    deepseek_agent.print_response(deepseek_prompt, stream=True)\n```\n\n## Run the Example\n\n<Steps>\n  <Snippet file=\"create-venv-step.mdx\" />\n\n  <Step title=\"Install dependencies\">\n    ```bash\n    uv pip install -U agno openai\n    ```\n  </Step>\n\n  <Step title=\"Export your API keys\">\n    <CodeGroup>\n    ```bash Mac/Linux\n    export DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    export OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n\n    ```bash Windows\n    $Env:DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    $Env:OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n    </CodeGroup>\n  </Step>\n\n  <Step title=\"Run the example\">\n    Save the code above as `trolley_problem.py`, then run:\n    ```bash\n    python trolley_problem.py\n    ```\n  </Step>\n</Steps>\n\nFull source: [cookbook/10_reasoning/agents/trolley_problem.py](https://github.com/agno-agi/agno/blob/main/cookbook/10_reasoning/agents/trolley_problem.py)\n"}, "2f2cceffa84d4e0c2e6fca0e69ca010b6c29d67e": {"t": 1785764993.98493, "n": 3026, "h": "bdc32281ea54a6e2", "c": "---\ntitle: \"Trolley Problem Analysis\"\ndescription: \"Demonstrates built-in and DeepSeek-backed reasoning for ethical analysis.\"\nsource: cookbook/10_reasoning/agents/trolley_problem.py\n---\n\n```python trolley_problem.py\n\"\"\"\nTrolley Problem Analysis\n========================\n\nDemonstrates built-in and DeepSeek-backed reasoning for ethical analysis.\n\"\"\"\n\nfrom agno.agent import Agent\nfrom agno.models.deepseek import DeepSeek\nfrom agno.models.openai import OpenAIChat\n\n# ---------------------------------------------------------------------------\n# Create Agents\n# ---------------------------------------------------------------------------\ncot_prompt = (\n    \"Solve the trolley problem. Evaluate multiple ethical frameworks. \"\n    \"Include an ASCII diagram of your solution.\"\n)\n\ndeepseek_prompt = (\n    \"You are a philosopher tasked with analyzing the classic 'Trolley Problem'. In this scenario, a runaway trolley \"\n    \"is barreling down the tracks towards five people who are tied up and unable to move. You are standing next to \"\n    \"a large stranger on a footbridge above the tracks. The only way to save the five people is to push this stranger \"\n    \"off the bridge onto the tracks below. This will kill the stranger, but save the five people on the tracks. \"\n    \"Should you push the stranger to save the five people? Provide a well-reasoned answer considering utilitarian, \"\n    \"deontological, and virtue ethics frameworks. \"\n    \"Include a simple ASCII art diagram to illustrate the scenario.\"\n)\n\ncot_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning=True,\n    markdown=True,\n)\n\ndeepseek_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning_model=DeepSeek(id=\"deepseek-reasoner\"),\n    markdown=True,\n)\n\n# ---------------------------------------------------------------------------\n# Run Agents\n# ---------------------------------------------------------------------------\nif __name__ == \"__main__\":\n    print(\"=== Built-in Chain Of Thought ===\")\n    cot_agent.print_response(cot_prompt, stream=True, show_full_reasoning=True)\n\n    print(\"\\n=== DeepSeek Reasoning Model ===\")\n    deepseek_agent.print_response(deepseek_prompt, stream=True)\n```\n\n## Run the Example\n\n<Steps>\n  <Snippet file=\"create-venv-step.mdx\" />\n\n  <Step title=\"Install dependencies\">\n    ```bash\n    uv pip install -U agno openai\n    ```\n  </Step>\n\n  <Step title=\"Export your API keys\">\n    <CodeGroup>\n    ```bash Mac/Linux\n    export DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    export OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n\n    ```bash Windows\n    $Env:DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    $Env:OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n    </CodeGroup>\n  </Step>\n\n  <Step title=\"Run the example\">\n    Save the code above as `trolley_problem.py`, then run:\n    ```bash\n    python trolley_problem.py\n    ```\n  </Step>\n</Steps>\n\nFull source: [cookbook/10_reasoning/agents/trolley_problem.py](https://github.com/agno-agi/agno/blob/main/cookbook/10_reasoning/agents/trolley_problem.py)\n"}, "b635e4aba367b4e756cc95854f6313a093210041": {"t": 1785764995.4520795, "n": 2366, "h": "47bffe8d5c31a604", "c": "---\ntitle: \"Trolley Problem Analysis\"\ndescription: \"Demonstrates built-in and DeepSeek-backed reasoning for ethical analysis.\"\n---\n```python\n\"\"\"\nTrolley Problem Analysis\n========================\n\nDemonstrates built-in and DeepSeek-backed reasoning for ethical analysis.\n\"\"\"\n\nfrom agno.agent import Agent\nfrom agno.models.deepseek import DeepSeek\nfrom agno.models.openai import OpenAIChat\n\n# ---------------------------------------------------------------------------\n# Create Agents\n# ---------------------------------------------------------------------------\ncot_prompt = (\n    \"Solve the trolley problem. Evaluate multiple ethical frameworks. \"\n    \"Include an ASCII diagram of your solution.\"\n)\n\ndeepseek_prompt = (\n    \"You are a philosopher tasked with analyzing the classic 'Trolley Problem'. In this scenario, a runaway trolley \"\n    \"is barreling down the tracks towards five people who are tied up and unable to move. You are standing next to \"\n    \"a large stranger on a footbridge above the tracks. The only way to save the five people is to push this stranger \"\n    \"off the bridge onto the tracks below. This will kill the stranger, but save the five people on the tracks. \"\n    \"Should you push the stranger to save the five people? Provide a well-reasoned answer considering utilitarian, \"\n    \"deontological, and virtue ethics frameworks. \"\n    \"Include a simple ASCII art diagram to illustrate the scenario.\"\n)\n\ncot_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning=True,\n    markdown=True,\n)\n\ndeepseek_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning_model=DeepSeek(id=\"deepseek-reasoner\"),\n    markdown=True,\n)\n\n# ---------------------------------------------------------------------------\n# Run Agents\n# ---------------------------------------------------------------------------\nif __name__ == \"__main__\":\n    print(\"=== Built-in Chain Of Thought ===\")\n    cot_agent.print_response(cot_prompt, stream=True, show_full_reasoning=True)\n\n    print(\"\\n=== DeepSeek Reasoning Model ===\")\n    deepseek_agent.print_response(deepseek_prompt, stream=True)\n```\n\n## Run the Example\n```bash\n# Clone and setup repo\ngit clone https://github.com/agno-agi/agno.git\ncd agno/cookbook/10_reasoning/agents\n\n# Create and activate virtual environment\n./scripts/demo_setup.sh\nsource .venvs/demo/bin/activate\n\npython trolley_problem.py\n```\n"}, "ef8fc4de8c409415dd1573552deaa335520980de": {"t": 1786068464.7141001, "n": 3151, "h": "c9822f25d9a3ef9a", "c": "---\ntitle: \"Trolley Problem Analysis\"\ndescription: \"Solve the trolley problem with ASCII diagrams using a built-in chain-of-thought agent and a deepseek-reasoner-backed agent.\"\nsource: cookbook/10_reasoning/agents/trolley_problem.py\n---\n\nDemonstrates built-in and DeepSeek-backed reasoning for ethical analysis.\n\n```python trolley_problem.py\n\"\"\"\nTrolley Problem Analysis\n========================\n\nDemonstrates built-in and DeepSeek-backed reasoning for ethical analysis.\n\"\"\"\n\nfrom agno.agent import Agent\nfrom agno.models.deepseek import DeepSeek\nfrom agno.models.openai import OpenAIChat\n\n# ---------------------------------------------------------------------------\n# Create Agents\n# ---------------------------------------------------------------------------\ncot_prompt = (\n    \"Solve the trolley problem. Evaluate multiple ethical frameworks. \"\n    \"Include an ASCII diagram of your solution.\"\n)\n\ndeepseek_prompt = (\n    \"You are a philosopher tasked with analyzing the classic 'Trolley Problem'. In this scenario, a runaway trolley \"\n    \"is barreling down the tracks towards five people who are tied up and unable to move. You are standing next to \"\n    \"a large stranger on a footbridge above the tracks. The only way to save the five people is to push this stranger \"\n    \"off the bridge onto the tracks below. This will kill the stranger, but save the five people on the tracks. \"\n    \"Should you push the stranger to save the five people? Provide a well-reasoned answer considering utilitarian, \"\n    \"deontological, and virtue ethics frameworks. \"\n    \"Include a simple ASCII art diagram to illustrate the scenario.\"\n)\n\ncot_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning=True,\n    markdown=True,\n)\n\ndeepseek_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning_model=DeepSeek(id=\"deepseek-reasoner\"),\n    markdown=True,\n)\n\n# ---------------------------------------------------------------------------\n# Run Agents\n# ---------------------------------------------------------------------------\nif __name__ == \"__main__\":\n    print(\"=== Built-in Chain Of Thought ===\")\n    cot_agent.print_response(cot_prompt, stream=True, show_full_reasoning=True)\n\n    print(\"\\n=== DeepSeek Reasoning Model ===\")\n    deepseek_agent.print_response(deepseek_prompt, stream=True)\n```\n\n## Run the Example\n\n<Steps>\n  <Snippet file=\"create-venv-step.mdx\" />\n\n  <Step title=\"Install dependencies\">\n    ```bash\n    uv pip install -U agno openai\n    ```\n  </Step>\n\n  <Step title=\"Export your API keys\">\n    <CodeGroup>\n    ```bash Mac/Linux\n    export DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    export OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n\n    ```bash Windows\n    $Env:DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    $Env:OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n    </CodeGroup>\n  </Step>\n\n  <Step title=\"Run the example\">\n    Save the code above as `trolley_problem.py`, then run:\n    ```bash\n    python trolley_problem.py\n    ```\n  </Step>\n</Steps>\n\nFull source: [cookbook/10_reasoning/agents/trolley_problem.py](https://github.com/agno-agi/agno/blob/main/cookbook/10_reasoning/agents/trolley_problem.py)\n"}}, "examples/reasoning/agents/scientific-research.mdx": {"2c5a11695c8a13de4b8cf1be164317854a9642de": {"t": 1785746505.656262, "n": 3020, "h": "1a0f5674066a5dbb", "c": "---\ntitle: \"Scientific Abstract Critique\"\ndescription: \"Critique a small-sample teaching-method abstract for bias and flawed conclusions, comparing gpt-4o built-in chain-of-thought with a deepseek-reasoner reasoning model.\"\nsource: cookbook/10_reasoning/agents/scientific_research.py\n---\n\nDemonstrates built-in and DeepSeek-backed reasoning for methodology critique.\n\n```python scientific_research.py\n\"\"\"\nScientific Abstract Critique\n============================\n\nDemonstrates built-in and DeepSeek-backed reasoning for methodology critique.\n\"\"\"\n\nfrom agno.agent import Agent\nfrom agno.models.deepseek import DeepSeek\nfrom agno.models.openai import OpenAIChat\n\n# ---------------------------------------------------------------------------\n# Create Agents\n# ---------------------------------------------------------------------------\ntask = (\n    \"Read the following abstract of a scientific paper and provide a critical evaluation of its methodology,\"\n    \"results, conclusions, and any potential biases or flaws:\\n\\n\"\n    \"Abstract: This study examines the effect of a new teaching method on student performance in mathematics. \"\n    \"A sample of 30 students was selected from a single school and taught using the new method over one semester. \"\n    \"The results showed a 15% increase in test scores compared to the previous semester. \"\n    \"The study concludes that the new teaching method is effective in improving mathematical performance among high school students.\"\n)\n\ncot_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning=True,\n    markdown=True,\n)\n\ndeepseek_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning_model=DeepSeek(id=\"deepseek-reasoner\"),\n    markdown=True,\n)\n\n# ---------------------------------------------------------------------------\n# Run Agents\n# ---------------------------------------------------------------------------\nif __name__ == \"__main__\":\n    print(\"=== Built-in Chain Of Thought ===\")\n    cot_agent.print_response(task, stream=True, show_full_reasoning=True)\n\n    print(\"\\n=== DeepSeek Reasoning Model ===\")\n    deepseek_agent.print_response(task, stream=True, show_full_reasoning=True)\n```\n\n## Run the Example\n\n<Steps>\n  <Snippet file=\"create-venv-step.mdx\" />\n\n  <Step title=\"Install dependencies\">\n    ```bash\n    uv pip install -U agno openai\n    ```\n  </Step>\n\n  <Step title=\"Export your API keys\">\n    <CodeGroup>\n    ```bash Mac/Linux\n    export DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    export OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n\n    ```bash Windows\n    $Env:DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    $Env:OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n    </CodeGroup>\n  </Step>\n\n  <Step title=\"Run the example\">\n    Save the code above as `scientific_research.py`, then run:\n    ```bash\n    python scientific_research.py\n    ```\n  </Step>\n</Steps>\n\nFull source: [cookbook/10_reasoning/agents/scientific_research.py](https://github.com/agno-agi/agno/blob/main/cookbook/10_reasoning/agents/scientific_research.py)\n"}, "6917c89afd1df7aa2a58b9b9c63f9c8cc19a354e": {"t": 1785764992.526958, "n": 3020, "h": "1a0f5674066a5dbb", "c": "---\ntitle: \"Scientific Abstract Critique\"\ndescription: \"Critique a small-sample teaching-method abstract for bias and flawed conclusions, comparing gpt-4o built-in chain-of-thought with a deepseek-reasoner reasoning model.\"\nsource: cookbook/10_reasoning/agents/scientific_research.py\n---\n\nDemonstrates built-in and DeepSeek-backed reasoning for methodology critique.\n\n```python scientific_research.py\n\"\"\"\nScientific Abstract Critique\n============================\n\nDemonstrates built-in and DeepSeek-backed reasoning for methodology critique.\n\"\"\"\n\nfrom agno.agent import Agent\nfrom agno.models.deepseek import DeepSeek\nfrom agno.models.openai import OpenAIChat\n\n# ---------------------------------------------------------------------------\n# Create Agents\n# ---------------------------------------------------------------------------\ntask = (\n    \"Read the following abstract of a scientific paper and provide a critical evaluation of its methodology,\"\n    \"results, conclusions, and any potential biases or flaws:\\n\\n\"\n    \"Abstract: This study examines the effect of a new teaching method on student performance in mathematics. \"\n    \"A sample of 30 students was selected from a single school and taught using the new method over one semester. \"\n    \"The results showed a 15% increase in test scores compared to the previous semester. \"\n    \"The study concludes that the new teaching method is effective in improving mathematical performance among high school students.\"\n)\n\ncot_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning=True,\n    markdown=True,\n)\n\ndeepseek_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning_model=DeepSeek(id=\"deepseek-reasoner\"),\n    markdown=True,\n)\n\n# ---------------------------------------------------------------------------\n# Run Agents\n# ---------------------------------------------------------------------------\nif __name__ == \"__main__\":\n    print(\"=== Built-in Chain Of Thought ===\")\n    cot_agent.print_response(task, stream=True, show_full_reasoning=True)\n\n    print(\"\\n=== DeepSeek Reasoning Model ===\")\n    deepseek_agent.print_response(task, stream=True, show_full_reasoning=True)\n```\n\n## Run the Example\n\n<Steps>\n  <Snippet file=\"create-venv-step.mdx\" />\n\n  <Step title=\"Install dependencies\">\n    ```bash\n    uv pip install -U agno openai\n    ```\n  </Step>\n\n  <Step title=\"Export your API keys\">\n    <CodeGroup>\n    ```bash Mac/Linux\n    export DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    export OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n\n    ```bash Windows\n    $Env:DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    $Env:OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n    </CodeGroup>\n  </Step>\n\n  <Step title=\"Run the example\">\n    Save the code above as `scientific_research.py`, then run:\n    ```bash\n    python scientific_research.py\n    ```\n  </Step>\n</Steps>\n\nFull source: [cookbook/10_reasoning/agents/scientific_research.py](https://github.com/agno-agi/agno/blob/main/cookbook/10_reasoning/agents/scientific_research.py)\n"}, "2f2cceffa84d4e0c2e6fca0e69ca010b6c29d67e": {"t": 1785764993.9534636, "n": 2852, "h": "f5253da0b883c4f0", "c": "---\ntitle: \"Scientific Abstract Critique\"\ndescription: \"Demonstrates built-in and DeepSeek-backed reasoning for methodology critique.\"\nsource: cookbook/10_reasoning/agents/scientific_research.py\n---\n\n```python scientific_research.py\n\"\"\"\nScientific Abstract Critique\n============================\n\nDemonstrates built-in and DeepSeek-backed reasoning for methodology critique.\n\"\"\"\n\nfrom agno.agent import Agent\nfrom agno.models.deepseek import DeepSeek\nfrom agno.models.openai import OpenAIChat\n\n# ---------------------------------------------------------------------------\n# Create Agents\n# ---------------------------------------------------------------------------\ntask = (\n    \"Read the following abstract of a scientific paper and provide a critical evaluation of its methodology,\"\n    \"results, conclusions, and any potential biases or flaws:\\n\\n\"\n    \"Abstract: This study examines the effect of a new teaching method on student performance in mathematics. \"\n    \"A sample of 30 students was selected from a single school and taught using the new method over one semester. \"\n    \"The results showed a 15% increase in test scores compared to the previous semester. \"\n    \"The study concludes that the new teaching method is effective in improving mathematical performance among high school students.\"\n)\n\ncot_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning=True,\n    markdown=True,\n)\n\ndeepseek_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning_model=DeepSeek(id=\"deepseek-reasoner\"),\n    markdown=True,\n)\n\n# ---------------------------------------------------------------------------\n# Run Agents\n# ---------------------------------------------------------------------------\nif __name__ == \"__main__\":\n    print(\"=== Built-in Chain Of Thought ===\")\n    cot_agent.print_response(task, stream=True, show_full_reasoning=True)\n\n    print(\"\\n=== DeepSeek Reasoning Model ===\")\n    deepseek_agent.print_response(task, stream=True, show_full_reasoning=True)\n```\n\n## Run the Example\n\n<Steps>\n  <Snippet file=\"create-venv-step.mdx\" />\n\n  <Step title=\"Install dependencies\">\n    ```bash\n    uv pip install -U agno openai\n    ```\n  </Step>\n\n  <Step title=\"Export your API keys\">\n    <CodeGroup>\n    ```bash Mac/Linux\n    export DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    export OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n\n    ```bash Windows\n    $Env:DEEPSEEK_API_KEY=\"your_deepseek_api_key_here\"\n    $Env:OPENAI_API_KEY=\"your_openai_api_key_here\"\n    ```\n    </CodeGroup>\n  </Step>\n\n  <Step title=\"Run the example\">\n    Save the code above as `scientific_research.py`, then run:\n    ```bash\n    python scientific_research.py\n    ```\n  </Step>\n</Steps>\n\nFull source: [cookbook/10_reasoning/agents/scientific_research.py](https://github.com/agno-agi/agno/blob/main/cookbook/10_reasoning/agents/scientific_research.py)\n"}, "b635e4aba367b4e756cc95854f6313a093210041": {"t": 1785764995.3619556, "n": 2172, "h": "6d63f111ffcb8f13", "c": "---\ntitle: \"Scientific Abstract Critique\"\ndescription: \"Demonstrates built-in and DeepSeek-backed reasoning for methodology critique.\"\n---\n```python\n\"\"\"\nScientific Abstract Critique\n============================\n\nDemonstrates built-in and DeepSeek-backed reasoning for methodology critique.\n\"\"\"\n\nfrom agno.agent import Agent\nfrom agno.models.deepseek import DeepSeek\nfrom agno.models.openai import OpenAIChat\n\n# ---------------------------------------------------------------------------\n# Create Agents\n# ---------------------------------------------------------------------------\ntask = (\n    \"Read the following abstract of a scientific paper and provide a critical evaluation of its methodology,\"\n    \"results, conclusions, and any potential biases or flaws:\\n\\n\"\n    \"Abstract: This study examines the effect of a new teaching method on student performance in mathematics. \"\n    \"A sample of 30 students was selected from a single school and taught using the new method over one semester. \"\n    \"The results showed a 15% increase in test scores compared to the previous semester. \"\n    \"The study concludes that the new teaching method is effective in improving mathematical performance among high school students.\"\n)\n\ncot_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning=True,\n    markdown=True,\n)\n\ndeepseek_agent = Agent(\n    model=OpenAIChat(id=\"gpt-4o\"),\n    reasoning_model=DeepSeek(id=\"deepseek-reasoner\"),\n    markdown=True,\n)\n\n# ---------------------------------------------------------------------------\n# Run Agents\n# ---------------------------------------------------------------------------\nif __name__ == \"__main__\":\n    print(\"=== Built-in Chain Of Thought ===\")\n    cot_agent.print_response(task, stream=True, show_full_reasoning=True)\n\n    print(\"\\n=== DeepSeek Reasoning Model ===\")\n    deepseek_agent.print_response(task, stream=True, show_full_reasoning=True)\n```\n\n## Run the Example\n```bash\n# Clone and setup repo\ngit clone https://github.com/agno-agi/agno.git\ncd agno/cookbook/10_reasoning/agents\n\n# Create and activate virtual environment\n./scripts/demo_setup.sh\nsource .venvs/demo/bin/activate\n\npython scientific_research.py\n```\n"}}, "scripts/examples_sync/description-overrides.json": {"ef8fc4de8c409415dd1573552deaa335520980de": {"t": 1786068456.3530712, "n": 175514, "h": "4cadf09ddc385f59", "c": "{\n  \"examples/agent-os/factories/workflow/tiered-workflow-factory\": \"Build subscription-specific workflow pipelines with a WorkflowFactory.\",\n  \"examples/agent-os/knowledge/agentos-docling-markdown-analyst\": \"Serve a Docling-backed markdown knowledge agent with PgVector through AgentOS.\",\n  \"examples/agents/state-and-session/dynamic-session-state\": \"Update session state inside a tool hook during an agent run.\",\n  \"examples/evals/performance/simple-response\": \"Measure baseline response latency for a single GPT-5.2 prompt.\",\n  \"examples/evals/reliability/team/ai-news\": \"Check whether a news-research team makes the expected delegation and web search tool calls.\",\n  \"examples/teams/basics/broadcast-mode\": \"Delegate the same task to every team member with TeamMode.broadcast.\",\n  \"examples/tools/tool-hooks/tool-hook\": \"Run logic before and after tool execution with a tool hook.\",\n  \"examples/workflows/advanced-concepts/nested-workflows/deeply-nested-workflow\": \"Compose a three-level workflow with parallel research branches and nested mini-workflows.\",\n  \"examples/agent-os/advanced-demo/agents\": \"Two Postgres-backed demo agents: Sage with Exa and web search, and Agno Assist with PgVector knowledge.\",\n  \"examples/agent-os/advanced-demo/checkpointing\": \"Enable tool-batch checkpointing on a research agent served through AgentOS with Postgres.\",\n  \"examples/agent-os/advanced-demo/file-output\": \"Serve an agent that generates downloadable files with FileGenerationTools in AgentOS.\",\n  \"examples/agent-os/advanced-demo/multiple-knowledge-bases\": \"Serve a PgVector-backed knowledge agent and manage documents through AgentOS knowledge endpoints.\",\n  \"examples/agent-os/advanced-demo/reasoning-demo\": \"Run chain-of-thought, reasoning-model, and ReasoningTools agents plus a finance team in one AgentOS.\",\n  \"examples/agent-os/advanced-demo/teams\": \"Build a finance team of web search and YFinance agents led by Claude with ReasoningTools.\",\n  \"examples/agent-os/advanced-demo/teams-demo\": \"Serve research, multimodal, and financial news teams with file, audio, and video agents in AgentOS.\",\n  \"examples/agent-os/background-tasks/background-evals-example\": \"Run one AgentAsJudgeEval post-hook synchronously and another in the background in AgentOS.\",\n  \"examples/agent-os/basic\": \"Serve a minimal AgentOS with one Postgres-backed agent, team, and workflow.\",\n  \"examples/agent-os/client/server\": \"Start the AgentOS server that backs the client examples: agents, a team, a workflow, and knowledge.\",\n  \"examples/agent-os/customize/update-from-lifespan\": \"Register a second agent and resync AgentOS from a FastAPI lifespan function.\",\n  \"examples/agent-os/dbs/surreal-db/agents\": \"Define Agno Assist with agentic memory, the Agno MCP server, and SurrealDB session storage.\",\n  \"examples/agent-os/dbs/surreal-db/db\": \"Configure a SurrealDb instance with connection URL, credentials, namespace, and database.\",\n  \"examples/agent-os/dbs/surreal-db/run\": \"Serve the SurrealDB-backed agent, team, and workflow together in one AgentOS.\",\n  \"examples/agent-os/dbs/surreal-db/teams\": \"Build a reasoning finance team with web search and YFinance agents backed by SurrealDB.\",\n  \"examples/agent-os/dbs/surreal-db/workflows\": \"Define a SurrealDB-backed research workflow with a team step, writer step, and Pydantic input schema.\",\n  \"examples/agent-os/followup/followups-agentos\": \"Enable built-in followup suggestions on an agent and team served through AgentOS.\",\n  \"examples/agent-os/human-in-the-loop/workflow/main\": \"Serve every workflow human-in-the-loop example from this directory in a single AgentOS.\",\n  \"examples/agent-os/human-in-the-loop/workflow/workflow-db\": \"Shared Postgres database used by the workflow human-in-the-loop examples.\",\n  \"examples/agent-os/interfaces/a2a/agent-with-tools\": \"Expose a web search agent over the A2A protocol with message:send and message:stream.\",\n  \"examples/agent-os/interfaces/a2a/basic\": \"Serve a minimal chat agent over the A2A protocol with a2a_interface enabled.\",\n  \"examples/agent-os/interfaces/a2a/multi-agent-a2a/airbnb-agent\": \"Serve an Airbnb search agent built on the OpenBNB MCP server over the A2A protocol.\",\n  \"examples/agent-os/interfaces/a2a/multi-agent-a2a/trip-planning-a2a-client\": \"Orchestrate trip planning by calling remote Airbnb and weather agents through A2A tool functions.\",\n  \"examples/agent-os/interfaces/a2a/multi-agent-a2a/weather-agent\": \"Serve a weather reporter agent with OpenWeatherTools over the A2A protocol.\",\n  \"examples/agent-os/interfaces/a2a/reasoning-agent\": \"Expose an o4-mini reasoning agent with web search over the A2A protocol.\",\n  \"examples/agent-os/interfaces/a2a/research-team\": \"Expose a researcher and writer team over the A2A protocol from AgentOS.\",\n  \"examples/agent-os/interfaces/a2a/structured-output\": \"Return a MovieScript Pydantic schema from an agent served over the A2A protocol.\",\n  \"examples/agent-os/interfaces/agui/agent-with-tools\": \"Combine server-side web search with a frontend haiku tool in an agent served over AG-UI.\",\n  \"examples/agent-os/interfaces/agui/basic\": \"Serve a minimal assistant agent through the AG-UI interface in AgentOS.\",\n  \"examples/agent-os/interfaces/agui/multiple-instances\": \"Mount two agents on separate AG-UI endpoints with the prefix parameter.\",\n  \"examples/agent-os/interfaces/agui/reasoning-agent\": \"Serve an o4-mini reasoning agent with web search through the AG-UI interface.\",\n  \"examples/agent-os/interfaces/agui/research-team\": \"Serve a researcher and writer team through the AG-UI interface in AgentOS.\",\n  \"examples/agent-os/interfaces/agui/structured-output\": \"Return a MovieScript Pydantic schema from an agent served over AG-UI.\",\n  \"examples/agent-os/interfaces/all-interfaces\": \"Run Slack, Telegram, WhatsApp, AG-UI, and A2A interfaces together in a single AgentOS.\",\n  \"examples/agent-os/interfaces/whatsapp/agent-with-media\": \"Handle media messages on WhatsApp with a Gemini agent and SQLite session history.\",\n  \"examples/agent-os/interfaces/whatsapp/agent-with-user-memory\": \"Capture user memories on WhatsApp with a Gemini agent, MemoryManager, and agentic memory.\",\n  \"examples/agent-os/interfaces/whatsapp/basic\": \"Serve a WhatsApp chat agent that replies in short conversational paragraphs with SQLite history.\",\n  \"examples/agent-os/interfaces/whatsapp/image-generation-model\": \"Generate images on WhatsApp with a Gemini model that returns text and image modalities.\",\n  \"examples/agent-os/interfaces/whatsapp/reasoning-agent\": \"Run a Claude finance agent with ReasoningTools and YFinance on WhatsApp with visible reasoning.\",\n  \"examples/agent-os/knowledge/agentos-excel-analyst\": \"Query Excel spreadsheet data loaded into PgVector knowledge through an AgentOS agent.\",\n  \"examples/agent-os/knowledge/agno-docs-agent\": \"Serve a docs-answering agent with PgVector hybrid search knowledge over the Agno docs in AgentOS.\",\n  \"examples/agent-os/learnings/rest-api-learnings\": \"Create, list, update, and delete learning records through the AgentOS /learnings REST endpoints.\",\n  \"examples/agent-os/os-config/basic\": \"Configure AgentOS in code with AgentOSConfig: manifest quick prompts and memory database display names.\",\n  \"examples/agent-os/os-config/yaml-config\": \"Load AgentOS configuration from a YAML file and attach Slack and WhatsApp interfaces.\",\n  \"examples/agent-os/rbac/test-scopes\": \"Test RBAC scope enforcement with JWT tokens against agent, workflow, and component endpoints.\",\n  \"examples/agent-os/remote/remote-agent\": \"Call an agent hosted on another AgentOS with RemoteAgent, in both single-shot and streaming mode.\",\n  \"examples/agent-os/remote/remote-team\": \"Run a team hosted on another AgentOS with RemoteTeam, with and without streaming.\",\n  \"examples/agent-os/remote/server\": \"Start an AgentOS server with two agents, a team, a QA workflow, and ChromaDb knowledge for client examples.\",\n  \"examples/agent-os/scheduler/rest-api-schedules\": \"Create, update, trigger, and delete cron schedules through the AgentOS /schedules REST endpoints.\",\n  \"examples/agent-os/scheduler/run-history\": \"Inspect schedule run history with SchedulerConsole, using simulated success and failure records.\",\n  \"examples/agent-os/scheduler/schedule-validation\": \"Catch invalid cron, timezone, and duplicate-name errors when creating schedules with ScheduleManager.\",\n  \"examples/agent-os/scheduler/scheduler-with-agentos\": \"Enable cron polling with scheduler=True on AgentOS and create schedules through the REST API.\",\n  \"examples/agent-os/scheduler/team-workflow-schedules\": \"Schedule team and workflow runs by cron, plus a GET health-check schedule, with ScheduleManager.\",\n  \"examples/agent-os/skills/sample-skills/system-info/scripts/get-system-info\": \"Print OS, Python version, and host details as JSON from a system-info skill script.\",\n  \"examples/agent-os/skills/sample-skills/system-info/scripts/list-directory\": \"List directory entries with type and size as JSON from a system-info skill script.\",\n  \"examples/agent-os/skills/skills-with-agentos\": \"Give an agent local skill scripts with LocalSkills and serve it on AgentOS.\",\n  \"examples/agent-os/tracing/agent-with-knowledge-tracing\": \"Trace a knowledge agent's runs, including PgVector searches, with tracing enabled on AgentOS.\",\n  \"examples/agent-os/tracing/agent-with-reasoning-tools-tracing\": \"Capture reasoning steps in traces by serving a ReasoningTools agent with tracing enabled.\",\n  \"examples/agent-os/tracing/basic-agent-tracing\": \"Enable tracing on AgentOS to record every run of a HackerNews agent in SQLite.\",\n  \"examples/agent-os/tracing/basic-team-tracing\": \"Enable tracing for a team; AgentOS traces the team and its members without per-agent setup.\",\n  \"examples/agent-os/tracing/basic-workflow-tracing\": \"Trace a research workflow with a conditional fact-check step by enabling tracing on AgentOS.\",\n  \"examples/agent-os/tracing/dbs/basic-agent-with-mongodb\": \"Store agent traces in MongoDB by pairing MongoDb with tracing=True on AgentOS.\",\n  \"examples/agent-os/tracing/dbs/basic-agent-with-postgresdb\": \"Store agent traces in Postgres by pairing PostgresDb with tracing=True on AgentOS.\",\n  \"examples/agent-os/tracing/dbs/basic-agent-with-sqlite\": \"Store agent traces in SQLite by pairing SqliteDb with tracing=True on AgentOS.\",\n  \"examples/agent-os/tracing/tracing-with-multi-db-and-tracing-flag\": \"Route traces to a dedicated database when each agent has its own db, using tracing=True on AgentOS.\",\n  \"examples/agent-os/tracing/tracing-with-multi-db-scenario\": \"Configure tracing manually with setup_tracing, batching exports to a dedicated traces database.\",\n  \"examples/agent-os/workflow/basic-workflow\": \"Serve a two-step research and content planning workflow on AgentOS.\",\n  \"examples/agent-os/workflow/basic-workflow-team\": \"Serve a workflow whose research step runs a two-agent team before a content planning step.\",\n  \"examples/agent-os/workflow/customer-research-workflow-parallel\": \"Run customer research steps in parallel with session state, then consolidate and recommend tasks.\",\n  \"examples/agent-os/workflow/workflow-with-conditional\": \"Add a Condition step that triggers fact-checking only when the summary contains factual claims.\",\n  \"examples/agent-os/workflow/workflow-with-custom-function-updating-session-state\": \"Use a custom function step that reads and updates workflow session state across planning runs.\",\n  \"examples/agent-os/workflow/workflow-with-history\": \"Give workflow steps conversation history with add_workflow_history_to_steps in a meal planner.\",\n  \"examples/agent-os/workflow/workflow-with-input-schema\": \"Validate workflow input against a Pydantic model with input_schema.\",\n  \"examples/agent-os/workflow/workflow-with-loop\": \"Repeat research steps in a Loop until an end condition passes or max_iterations is reached.\",\n  \"examples/agent-os/workflow/workflow-with-nested-steps\": \"Run a nested deep-research Loop through a Router before the report-writing step.\",\n  \"examples/agent-os/workflow/workflow-with-parallel\": \"Run two research steps in Parallel before sequential writing and review steps.\",\n  \"examples/agent-os/workflow/workflow-with-parallel-and-custom-function-step-stream\": \"Stream events from async custom-function steps running in Parallel inside a workflow.\",\n  \"examples/agent-os/workflow/workflow-with-router\": \"Route to HackerNews or web research based on topic keywords with a Router step.\",\n  \"examples/agent-os/workflow/workflow-with-steps\": \"Group research, writing, and editing into a reusable Steps sequence inside a workflow.\",\n  \"examples/agents/advanced/agent-serialization\": \"Serialize an agent with to_dict and from_dict, and persist versions with save and load.\",\n  \"examples/agents/advanced/debug\": \"Enable verbose debug output for every run with debug_mode, or turn it on for a single run.\",\n  \"examples/agents/advanced/interchange-model/claude-gemini\": \"Switch a single agent between Claude and Gemini mid-conversation with shared Postgres history.\",\n  \"examples/agents/advanced/interchange-model/openai-chat-responses\": \"Alternate one agent between OpenAIChat and OpenAIResponses mid-session over shared Postgres history.\",\n  \"examples/agents/advanced/interchange-model/openai-claude\": \"Swap a single agent between GPT-4o and Claude mid-session with tool-call history stored in Postgres.\",\n  \"examples/agents/advanced/interchange-model/openai-gemini\": \"Move one conversation between GPT-4o and Gemini across turns with history persisted in Postgres.\",\n  \"examples/agents/advanced/metrics\": \"Inspect per-message, run, and session metrics after a YFinance tool call on a Postgres-backed agent.\",\n  \"examples/agents/approvals/audit-approval-overview\": \"Compare pre-execution @approval records with audit-type approvals logged after the tool runs in SQLite.\",\n  \"examples/agents/context-management/instructions\": \"Add the current date and time to the agent's context with add_datetime_to_context and a timezone.\",\n  \"examples/agents/dependencies/dynamic-tools\": \"Build the tool list at runtime from a function that reads session state off the RunContext.\",\n  \"examples/agents/guardrails/custom-guardrail\": \"Subclass BaseGuardrail and raise InputCheckError to block security-abuse prompts as a pre-hook.\",\n  \"examples/agents/guardrails/output-guardrail\": \"Reject responses under 20 characters with a post-hook that raises OutputCheckError.\",\n  \"examples/agents/human-in-the-loop/agentic-user-input\": \"Let the agent request missing fields at runtime with UserControlFlowTools, then continue the run.\",\n  \"examples/agents/human-in-the-loop/confirmation-advanced\": \"Confirm or reject tool calls across a custom HackerNews tool and WikipediaTools in one run.\",\n  \"examples/agents/human-in-the-loop/confirmation-required-mcp-toolkit\": \"Require confirmation for an MCP server tool and resume an async streamed run after approval.\",\n  \"examples/agents/human-in-the-loop/confirmation-toolkit\": \"Gate a toolkit's web_search tool behind user confirmation with requires_confirmation_tools.\",\n  \"examples/agents/human-in-the-loop/external-tool-execution\": \"Pause the run for a shell tool marked external_execution, execute it yourself, and continue with the result.\",\n  \"examples/agents/human-in-the-loop/mixed-external-and-regular-tools\": \"Combine a regular tool with an external_execution tool; the agent auto-runs one and pauses for the other.\",\n  \"examples/agents/human-in-the-loop/user-feedback\": \"Pause a trip-planning agent with UserFeedbackTools and collect answers to multiple-choice questions.\",\n  \"examples/agents/human-in-the-loop/user-input-required\": \"Mark a tool with requires_user_input so the run pauses to collect the to_address field.\",\n  \"examples/agents/input-output/input-formats\": \"Send an OpenAI-style message dict with text and image_url parts as agent input.\",\n  \"examples/agents/input-output/input-schema\": \"Validate agent input against a Pydantic input_schema, passing either a dict or a model instance.\",\n  \"examples/agents/input-output/parser-model\": \"Pair output_schema with a parser_model that turns the response into a NationalParkAdventure object.\",\n  \"examples/agents/knowledge/agentic-rag\": \"Agentic RAG with PgVector hybrid search: the agent queries a recipe knowledge base on demand.\",\n  \"examples/agents/knowledge/agentic-rag-with-reranking\": \"Agentic RAG over LanceDB with hybrid search and a Cohere reranker ordering results.\",\n  \"examples/agents/knowledge/traditional-rag\": \"Traditional RAG that injects PgVector search results into the prompt instead of using a search tool.\",\n  \"examples/agents/memory-and-learning/learning-machine\": \"Attach a LearningMachine that builds an agentic user profile and recalls it in a new session.\",\n  \"examples/agents/multimodal/audio-streaming\": \"Stream pcm16 audio from gpt-audio and write the chunks to a WAV file while printing the transcript.\",\n  \"examples/agents/multimodal/audio-to-text\": \"Transcribe an MP3 conversation with Gemini, labeling each speaker in the output.\",\n  \"examples/agents/reasoning/basic-reasoning\": \"Enable reasoning=True with step bounds, then stream the full reasoning trace.\",\n  \"examples/agents/state-and-session/chat-history\": \"Persist chat history in Postgres and read it back with get_chat_history() between turns.\",\n  \"examples/integrations/rag/local-rag-langchain-qdrant\": \"Local RAG with Ollama answering from a LangChain retriever backed by Qdrant and FastEmbed embeddings.\",\n  \"examples/integrations/surrealdb/custom-memory-instructions\": \"Scope what MemoryManager captures with custom instructions, storing memories in SurrealDB.\",\n  \"examples/integrations/surrealdb/db-tools-control\": \"Control memory writes with MemoryManager's add_memories and update_memories flags, backed by SurrealDB.\",\n  \"examples/integrations/surrealdb/memory-creation\": \"Create user memories from raw text and message history with MemoryManager on SurrealDB.\",\n  \"examples/integrations/surrealdb/memory-search-surreal\": \"Search SurrealDB memories with last_n, first_n, and agentic retrieval methods.\",\n  \"examples/integrations/surrealdb/standalone-memory-surreal\": \"Add, delete, and replace user memories directly in SurrealDB with a standalone MemoryManager.\",\n  \"examples/knowledge/integrations/rag/local-rag-langchain-qdrant\": \"Local RAG with Ollama answering from a LangChain retriever backed by Qdrant and FastEmbed embeddings.\",\n  \"examples/models/aimlapi/basic\": \"Run an AIMLAPI agent four ways: sync, async, and streaming variants of each.\",\n  \"examples/models/aimlapi/image-agent\": \"Analyze an image from a URL with Llama 3.2 Vision through AIML API.\",\n  \"examples/models/aimlapi/image-agent-bytes\": \"Pass an image as raw bytes to a Llama vision model served through AIMLAPI.\",\n  \"examples/models/aimlapi/image-agent-with-memory\": \"Analyze an image with a Llama vision model on AIMLAPI, then answer a follow-up using chat history.\",\n  \"examples/models/aimlapi/structured-output\": \"Return a typed MovieScript from AIML API using JSON mode and a Pydantic output schema.\",\n  \"examples/models/aimlapi/tool-use\": \"Answer current-events questions with web search tools on an AIML API model, sync and async.\",\n  \"examples/models/anthropic/basic\": \"Run a minimal Claude agent in sync, async, and streaming modes.\",\n  \"examples/models/anthropic/basic-with-timeout\": \"Set a one-second request timeout on Claude with the timeout parameter.\",\n  \"examples/models/anthropic/code-execution\": \"Compute statistics with Claude's server-side code execution tool via a beta flag.\",\n  \"examples/models/anthropic/csv-input\": \"Pass a downloaded IMDB CSV to Claude as a File attachment for box office analysis.\",\n  \"examples/models/anthropic/db\": \"Persist Claude chat history in Postgres so follow-up questions keep context.\",\n  \"examples/models/anthropic/financial-analyst-thinking\": \"Analyze a stock portfolio with interleaved thinking, calculator, and YFinance tools.\",\n  \"examples/models/anthropic/image-input-bytes\": \"Send an image to Claude as raw bytes and search the web for related news.\",\n  \"examples/models/anthropic/image-input-url\": \"Pass an image URL to Claude and combine vision with web search tools.\",\n  \"examples/models/anthropic/knowledge\": \"Give Claude a PgVector knowledge base built from a recipe PDF with Azure OpenAI embeddings.\",\n  \"examples/models/anthropic/mcp-connector\": \"Connect Claude to the DeepWiki MCP server with the native MCP connector beta.\",\n  \"examples/models/anthropic/pdf-input-bytes\": \"Send a PDF to Claude as raw bytes and read citations from the run output.\",\n  \"examples/models/anthropic/pdf-input-local\": \"Attach a local PDF file to a Claude prompt and print the response citations.\",\n  \"examples/models/anthropic/pdf-input-url\": \"Summarize a PDF fetched from a URL by passing it to Claude as a File.\",\n  \"examples/models/anthropic/structured-output\": \"Generate a typed MovieScript from Claude with a Pydantic output schema.\",\n  \"examples/models/anthropic/thinking\": \"Enable Claude extended thinking with a token budget and stream the response.\",\n  \"examples/models/anthropic/tool-use\": \"Call web search tools from Claude in sync, streaming, and async modes.\",\n  \"examples/models/anthropic/web-fetch\": \"Fetch and summarize a web page with Anthropic's native web_fetch tool.\",\n  \"examples/models/anthropic/web-search\": \"Use Anthropic's native web_search tool and inspect search metrics on the run output.\",\n  \"examples/models/aws/bedrock/basic\": \"Run a minimal agent on AWS Bedrock in sync, async, and streaming modes.\",\n  \"examples/models/aws/bedrock/image-agent-bytes\": \"Send image bytes to Amazon Nova Pro on Bedrock and fetch related news with web search.\",\n  \"examples/models/aws/bedrock/pdf-agent-bytes\": \"Pass a PDF as bytes to Amazon Nova Pro on Bedrock and extract a recipe from it.\",\n  \"examples/models/aws/bedrock/structured-output\": \"Return a typed MovieScript from a Bedrock model with a Pydantic output schema.\",\n  \"examples/models/aws/bedrock/tool-use\": \"Call web search tools from a Bedrock model in sync, streaming, and async modes.\",\n  \"examples/models/aws/claude/basic\": \"Run Claude on AWS Bedrock through the aws.Claude model class in sync, async, and streaming modes.\",\n  \"examples/models/aws/claude/db\": \"Store session history for Claude on Bedrock in Postgres with add_history_to_context.\",\n  \"examples/models/aws/claude/image-agent\": \"Describe an image URL with Claude on Bedrock and enrich it with web search.\",\n  \"examples/models/aws/claude/knowledge\": \"Attach a PgVector knowledge base of recipes to Claude running on Bedrock.\",\n  \"examples/models/aws/claude/structured-output\": \"Generate a typed MovieScript from Claude on Bedrock with a Pydantic output schema.\",\n  \"examples/models/aws/claude/tool-use\": \"Call web search tools from Claude on Bedrock in sync, streaming, and async modes.\",\n  \"examples/models/azure/ai-foundry/basic\": \"Run a minimal Phi-4 agent on Azure AI Foundry in sync, async, and streaming modes.\",\n  \"examples/models/azure/ai-foundry/db\": \"Persist Azure AI Foundry sessions in Postgres and carry history across turns.\",\n  \"examples/models/azure/ai-foundry/demo-cohere\": \"Run Cohere Command R through Azure AI Foundry with a single prompt.\",\n  \"examples/models/azure/ai-foundry/demo-mistral\": \"Run Mistral Large through Azure AI Foundry with a single prompt.\",\n  \"examples/models/azure/ai-foundry/image-agent\": \"Migrate an Azure AI Foundry image-URL agent from retired Llama 3.2 Vision to Llama 4 Scout.\",\n  \"examples/models/azure/ai-foundry/image-agent-bytes\": \"Migrate an Azure AI Foundry image-bytes agent from retired Llama 3.2 Vision to Llama 4 Scout.\",\n  \"examples/models/azure/ai-foundry/knowledge\": \"Query a PgVector knowledge base from an Azure AI Foundry model with Azure OpenAI embeddings.\",\n  \"examples/models/azure/ai-foundry/structured-output\": \"Use Azure OpenAI for supported Pydantic structured output because the v2.7.2 Azure AI Foundry source fails before sending a request.\",\n  \"examples/models/azure/ai-foundry/tool-use\": \"Call WebSearchTools from a Cohere Command R agent on Azure AI Foundry.\",\n  \"examples/models/azure/claude/basic\": \"Run Claude Sonnet on Azure AI Foundry with sync, async, and streaming responses.\",\n  \"examples/models/azure/claude/thinking\": \"Enable extended thinking with a token budget on Claude via Azure AI Foundry.\",\n  \"examples/models/azure/claude/tool-use\": \"Add WebSearchTools to a Claude agent running on Azure AI Foundry.\",\n  \"examples/models/azure/openai/basic\": \"Create a minimal AzureOpenAI agent and print sync, async, and streamed responses.\",\n  \"examples/models/azure/openai/db\": \"Persist AzureOpenAI agent sessions in Postgres and add history to follow-up runs.\",\n  \"examples/models/azure/openai/knowledge\": \"Load a recipe PDF into PgVector with AzureOpenAIEmbedder and query it from an agent.\",\n  \"examples/models/azure/openai/structured-output\": \"Return a typed MovieScript from an AzureOpenAI agent with output_schema.\",\n  \"examples/models/azure/openai/tool-use\": \"Stream an AzureOpenAI agent's response as it calls WebSearchTools.\",\n  \"examples/models/cerebras-openai/basic\": \"Compare synchronous, asynchronous, and streaming calls through Cerebras' OpenAI-compatible endpoint.\",\n  \"examples/models/cerebras-openai/db\": \"Store CerebrasOpenAI agent sessions in Postgres with history added to context.\",\n  \"examples/models/cerebras-openai/knowledge\": \"Query a PgVector knowledge base built from a PDF with a CerebrasOpenAI agent.\",\n  \"examples/models/cerebras-openai/oss-gpt\": \"Use gpt-oss-120b on the Cerebras OpenAI-compatible endpoint with WebSearchTools.\",\n  \"examples/models/cerebras-openai/structured-output\": \"Return a typed MovieScript from a CerebrasOpenAI agent with output_schema.\",\n  \"examples/models/cerebras-openai/tool-use\": \"Call WebSearchTools from a CerebrasOpenAI agent in sync and async modes.\",\n  \"examples/models/cerebras/basic\": \"Compare synchronous, asynchronous, and streaming calls with the Cerebras model client.\",\n  \"examples/models/cerebras/db\": \"Store Cerebras agent sessions in Postgres so follow-up questions keep context.\",\n  \"examples/models/cerebras/knowledge\": \"Answer questions from a PDF knowledge base in PgVector with a Cerebras agent.\",\n  \"examples/models/cerebras/oss-gpt\": \"Run gpt-oss-120b on Cerebras with web search via WebSearchTools.\",\n  \"examples/models/cerebras/structured-output\": \"Compare strict and guided structured output on Cerebras with a MovieScript schema.\",\n  \"examples/models/cerebras/tool-use\": \"Call WebSearchTools from a Cerebras agent in synchronous and asynchronous modes.\",\n  \"examples/models/cloudflare/structured-output\": \"Compare JSON mode and native structured outputs on Cloudflare Workers AI.\",\n  \"examples/models/cohere/basic\": \"Run Cohere Command A with sync, async, and streaming responses.\",\n  \"examples/models/cohere/db\": \"Persist Cohere agent sessions in Postgres and reuse history across runs.\",\n  \"examples/models/cohere/image-agent\": \"Migrate the pinned Aya Vision 8B example to Command A Vision, then describe an image from a URL.\",\n  \"examples/models/cohere/image-agent-bytes\": \"Migrate the pinned Aya Vision 8B example to Command A Vision, then analyze raw image bytes.\",\n  \"examples/models/cohere/image-agent-local-file\": \"Migrate the pinned Aya Vision 8B example to Command A Vision, then analyze a local image with `Image(filepath=...)`.\",\n  \"examples/models/cohere/knowledge\": \"Query a recipe PDF stored in PgVector from a Cohere Command A agent.\",\n  \"examples/models/cohere/structured-output\": \"Generate a typed MovieScript with Cohere Command A and output_schema.\",\n  \"examples/models/cohere/tool-use\": \"Add WebSearchTools to a Cohere Command A agent for live news queries.\",\n  \"examples/models/cometapi/basic\": \"Run a GPT model through the CometAPI gateway in sync, async, and streaming modes.\",\n  \"examples/models/cometapi/image-agent\": \"Stream an image description from a vision model through CometAPI.\",\n  \"examples/models/cometapi/image-agent-with-memory\": \"Analyze an image with CometAPI and recall it later from SQLite session history.\",\n  \"examples/models/cometapi/structured-output\": \"Produce a MovieScript from CometAPI using output_schema with JSON mode.\",\n  \"examples/models/cometapi/tool-use\": \"Call WebSearchTools through CometAPI for prices, weather, and news queries.\",\n  \"examples/models/dashscope/basic\": \"Run Qwen Plus on DashScope with sync, async, and streaming responses.\",\n  \"examples/models/dashscope/image-agent\": \"Analyze image URLs with a tool-capable Qwen3-VL model and enrich answers with web search.\",\n  \"examples/models/dashscope/image-agent-bytes\": \"Analyze an image from raw bytes with Qwen VL Plus and search the web for context.\",\n  \"examples/models/dashscope/structured-output\": \"Return a structured MovieScript from Qwen Plus on DashScope with output_schema.\",\n  \"examples/models/dashscope/thinking-agent\": \"Enable thinking on DashScope's QVQ Max to reason through an image problem.\",\n  \"examples/models/dashscope/tool-use\": \"Give a DashScope qwen-plus agent web search tools and stream sync and async responses.\",\n  \"examples/models/deepinfra/basic\": \"Run DeepSeek V3 on DeepInfra with synchronous, asynchronous, and streaming responses.\",\n  \"examples/models/deepinfra/json-output\": \"Generate a MovieScript Pydantic object from a DeepInfra phi-4 agent with output_schema.\",\n  \"examples/models/deepinfra/tool-use\": \"Add web search to a DeepSeek V3 agent on DeepInfra and stream synchronous and asynchronous responses.\",\n  \"examples/models/deepseek/basic\": \"Run a DeepSeek V4 Flash agent with sync, async, and streaming responses.\",\n  \"examples/models/deepseek/reasoning-agent\": \"Solve the missionaries and cannibals puzzle with DeepSeek V4 Pro reasoning.\",\n  \"examples/models/deepseek/structured-output\": \"Compare JSON mode and native structured output for MovieScript generation on DeepSeek.\",\n  \"examples/models/deepseek/thinking-tool-calls\": \"Combine DeepSeek thinking mode with web search tool calls and show full reasoning.\",\n  \"examples/models/deepseek/tool-use\": \"Answer news questions with a DeepSeek V4 Flash agent using web search tools.\",\n  \"examples/models/fireworks/basic\": \"Run Fireworks GPT OSS 120B with synchronous, asynchronous, and streaming responses.\",\n  \"examples/models/fireworks/structured-output\": \"Return a MovieScript Pydantic object from a Fireworks GPT OSS 120B agent.\",\n  \"examples/models/fireworks/tool-use\": \"Stream web search results from a Fireworks GPT OSS 120B agent.\",\n  \"examples/models/google/gemini/audio-input-bytes-content\": \"Pass a downloaded WAV file to Gemini as raw audio bytes for analysis.\",\n  \"examples/models/google/gemini/audio-input-file-upload\": \"Upload an MP3 through the Gemini Files API and reuse the remote file across runs.\",\n  \"examples/models/google/gemini/audio-input-local-file-upload\": \"Send a local MP3 to Gemini with Audio(filepath) and stream the analysis.\",\n  \"examples/models/google/gemini/basic\": \"Run a Gemini 3.5 Flash agent with sync, async, and streaming responses.\",\n  \"examples/models/google/gemini/csv-input\": \"Analyze an IMDB movie dataset by attaching a CSV file to a Gemini agent.\",\n  \"examples/models/google/gemini/db\": \"Persist Gemini agent sessions in Postgres and carry history across turns.\",\n  \"examples/models/google/gemini/file-search-advanced\": \"Manage multiple Gemini File Search stores with custom chunking and metadata filters.\",\n  \"examples/models/google/gemini/file-search-basic\": \"Create a Gemini File Search store, upload a document, and query it with citations.\",\n  \"examples/models/google/gemini/file-search-rag-pipeline\": \"Build an async RAG pipeline over a directory of files with Gemini File Search.\",\n  \"examples/models/google/gemini/gemini-2-to-3\": \"Migrate a Gemini 2.5 session to Gemini 3.1 Pro Preview by sharing history through SqliteDb.\",\n  \"examples/models/google/gemini/gemini-3-pro\": \"Migrate the pinned Gemini 3 Pro example to Gemini 3.1 Pro Preview before using web search and SQLite chat history.\",\n  \"examples/models/google/gemini/gemini-3-pro-thinking-level\": \"Migrate the pinned thinking-level example to Gemini 3.1 Pro Preview and stream the async response.\",\n  \"examples/models/google/gemini/image-editing\": \"Send an image to Gemini and get back an edited version using image response modalities.\",\n  \"examples/models/google/gemini/image-generation\": \"Generate an image with Gemini response modalities and open it with PIL.\",\n  \"examples/models/google/gemini/image-input\": \"Pass an image URL to a Gemini agent and fetch related news with web search.\",\n  \"examples/models/google/gemini/image-input-file-upload\": \"Upload an image through the Gemini Files API and combine it with web search.\",\n  \"examples/models/google/gemini/knowledge\": \"Query a PDF knowledge base stored in PgVector with Gemini embeddings.\",\n  \"examples/models/google/gemini/pdf-input-local\": \"Attach a local PDF to a Gemini agent and ask follow-up questions with history.\",\n  \"examples/models/google/gemini/pdf-input-url\": \"Summarize a PDF from a URL and ask follow-ups using InMemoryDb chat history.\",\n  \"examples/models/google/gemini/storage-and-memory\": \"Combine PgVector knowledge, Postgres memory, session summaries, and web search on Gemini.\",\n  \"examples/models/google/gemini/structured-output\": \"Return an event plan as a Pydantic schema with enums, formats, and nested objects.\",\n  \"examples/models/google/gemini/text-to-speech\": \"Generate speech with a Gemini TTS model and write the audio to a WAV file.\",\n  \"examples/models/google/gemini/thinking-agent\": \"Solve a logic puzzle with a Gemini thinking budget and thought summaries enabled.\",\n  \"examples/models/google/gemini/tool-use\": \"Add web search tools to a Gemini agent and run sync, streaming, and async.\",\n  \"examples/models/google/gemini/url-context\": \"Compare two recipe pages by enabling Gemini's url_context to fetch page content.\",\n  \"examples/models/google/gemini/vertexai-with-credentials\": \"Authenticate Gemini on Vertex AI with explicit service account credentials.\",\n  \"examples/models/google/gemini/video-input-bytes-content\": \"Download an MP4 and pass it to Gemini as raw video bytes for analysis.\",\n  \"examples/models/google/gemini/video-input-file-upload\": \"Upload a video through the Gemini Files API and poll until it is ready to query.\",\n  \"examples/models/google/gemini/video-input-local-file-upload\": \"Pass a local video file to Gemini and ask the agent to describe its contents.\",\n  \"examples/models/google/gemini/video-input-youtube\": \"Send a YouTube URL as video input to Gemini and have the agent describe it.\",\n  \"examples/models/groq/agent-team\": \"Coordinate web search and YFinance agents in a team led by Groq Llama 3.3.\",\n  \"examples/models/groq/basic\": \"Run a Groq Llama 3.3 agent in sync, async, streaming, and non-streaming modes.\",\n  \"examples/models/groq/browser-search\": \"Answer live questions with Groq's built-in browser_search tool on GPT-OSS 20B.\",\n  \"examples/models/groq/db\": \"Store Groq agent sessions in Postgres so follow-up questions keep context.\",\n  \"examples/models/groq/image-agent\": \"Describe an image from a URL using Llama 4 Scout's vision support on Groq.\",\n  \"examples/models/groq/knowledge\": \"Answer questions from a PDF knowledge base in PgVector with a Groq agent.\",\n  \"examples/models/groq/metrics\": \"Inspect token and timing metrics per message after a Groq agent run.\",\n  \"examples/models/groq/reasoning-agent\": \"Pair DeepSeek R1 Distill for reasoning with Llama 3.3 for the final answer.\",\n  \"examples/models/groq/reasoning/basic\": \"Run DeepSeek R1 Distill Llama on Groq with sync and streaming responses.\",\n  \"examples/models/groq/reasoning/demo-deepseek-qwen\": \"Enable reasoning on Qwen 2.5 32B with DeepSeek R1 Distill Qwen as the reasoning model.\",\n  \"examples/models/groq/reasoning/demo-qwen-2-5-32b\": \"Query a PgVector recipe knowledge base with Qwen 2.5 32B on Groq.\",\n  \"examples/models/groq/reasoning/finance-agent\": \"Write a stock report with DeepSeek R1 Distill on Groq and YFinance tools.\",\n  \"examples/models/groq/research-agent-exa\": \"Research a topic with Exa search on Groq and save the markdown report to a file.\",\n  \"examples/models/groq/research-agent-seltz\": \"Generate a referenced research report with Seltz search and a Groq agent.\",\n  \"examples/models/groq/structured-output\": \"Return a MovieScript Pydantic object from Groq using JSON mode.\",\n  \"examples/models/groq/tool-use\": \"Combine web search and Newspaper4k article extraction in a Groq research agent.\",\n  \"examples/models/groq/transcription-agent\": \"Give an agent GroqTools to transcribe a hosted WAV file to English.\",\n  \"examples/models/groq/translation-agent\": \"Transcribe French audio, translate it, and generate English speech with GroqTools.\",\n  \"examples/models/huggingface/basic\": \"Run a Hugging Face Mistral 7B agent in sync, async, and streaming modes.\",\n  \"examples/models/huggingface/llama-essay-writer\": \"Write a 300-word essay on a user topic with GPT-OSS 120B on Hugging Face.\",\n  \"examples/models/huggingface/tool-use\": \"Add web search tools to a GPT-OSS 120B agent on Hugging Face.\",\n  \"examples/models/ibm/watsonx/basic\": \"Run a watsonx Mistral Small agent in sync, async, and streaming modes.\",\n  \"examples/models/ibm/watsonx/db\": \"Persist watsonx agent sessions in Postgres with history added to context.\",\n  \"examples/models/ibm/watsonx/image-agent-bytes\": \"Pass an image as raw bytes to Llama 3.2 Vision on watsonx for analysis.\",\n  \"examples/models/ibm/watsonx/knowledge\": \"Search a PgVector knowledge base loaded from a PDF with a watsonx agent.\",\n  \"examples/models/ibm/watsonx/structured-output\": \"Return a MovieScript Pydantic object from a watsonx agent with output_schema.\",\n  \"examples/models/ibm/watsonx/tool-use\": \"Stream web search results through a watsonx agent, sync and async.\",\n  \"examples/models/inception/basic\": \"Run an Inception Mercury 2 agent in sync, async, and streaming modes.\",\n  \"examples/models/inception/structured-output\": \"Generate a MovieScript schema with Mercury 2 using JSON mode and native structured output.\",\n  \"examples/models/inception/tool-use\": \"Add web search tools to an Inception Mercury 2 agent and stream the response.\",\n  \"examples/models/langdb/agent\": \"Fetch NVDA and TSLA stock prices with YFinance tools through LangDB.\",\n  \"examples/models/langdb/basic\": \"Run a Llama 3.1 70B agent through the LangDB gateway with streaming.\",\n  \"examples/models/langdb/data-analyst\": \"Load an IMDB movie CSV into DuckDB and query it with a LangDB agent.\",\n  \"examples/models/langdb/finance-agent\": \"Summarize stock fundamentals with a YFinance analyst agent on LangDB.\",\n  \"examples/models/langdb/structured-output\": \"Generate MovieScript objects with LangDB via JSON mode and structured outputs.\",\n  \"examples/models/langdb/web-search\": \"Answer current-events questions with web search tools through LangDB.\",\n  \"examples/models/litellm-openai/basic\": \"Run GPT-4o through a LiteLLM proxy server with sync and streaming responses.\",\n  \"examples/models/litellm-openai/tool-use\": \"Give a GPT-4o agent web search tools through a LiteLLM proxy.\",\n  \"examples/models/litellm/audio-input-agent\": \"Pass an MP3 file as audio input to a gpt-audio model routed through LiteLLM.\",\n  \"examples/models/litellm/basic\": \"Run a Hugging Face Mistral model through LiteLLM with sync, async, and streaming calls.\",\n  \"examples/models/litellm/basic-gpt\": \"Call OpenAI's gpt-4o through the LiteLLM model class for a basic agent response.\",\n  \"examples/models/litellm/db\": \"Persist LiteLLM agent sessions in SQLite and answer follow-ups with chat history in context.\",\n  \"examples/models/litellm/image-agent\": \"Analyze an image from a URL with gpt-4o via LiteLLM and pull related news using web search.\",\n  \"examples/models/litellm/image-agent-bytes\": \"Send an image as raw bytes to gpt-4o via LiteLLM and fetch related news with web search.\",\n  \"examples/models/litellm/knowledge\": \"Answer questions from a PDF knowledge base stored in PgVector using a LiteLLM agent.\",\n  \"examples/models/litellm/memory\": \"Keep the last three runs in context and inspect stored session messages with a LiteLLM agent.\",\n  \"examples/models/litellm/metrics\": \"Inspect per-message and run-level token metrics from a LiteLLM agent using YFinance tools.\",\n  \"examples/models/litellm/pdf-input-bytes\": \"Pass a downloaded PDF as raw bytes to a LiteLLM agent and summarize its contents.\",\n  \"examples/models/litellm/pdf-input-local\": \"Attach a local PDF file to a LiteLLM agent and ask questions about a specific recipe.\",\n  \"examples/models/litellm/pdf-input-url\": \"Attach a PDF by URL to a LiteLLM agent and ask for a recipe from the document.\",\n  \"examples/models/litellm/structured-output\": \"Return a MovieScript Pydantic model via LiteLLM using JSON mode and native structured outputs.\",\n  \"examples/models/litellm/tool-use\": \"Call YFinance tools from a LiteLLM agent with sync, streaming, and async runs.\",\n  \"examples/models/llama-cpp/basic\": \"Run a local GGUF model with LlamaCpp and print sync and streamed responses.\",\n  \"examples/models/llama-cpp/structured-output\": \"Generate a MovieScript Pydantic object from a local LlamaCpp model with output_schema.\",\n  \"examples/models/llama-cpp/tool-use\": \"Give a local LlamaCpp model web search tools and stream the tool-assisted answer.\",\n  \"examples/models/lmstudio/basic\": \"Run a Qwen model served by LM Studio with sync and streamed responses.\",\n  \"examples/models/lmstudio/db\": \"Store LM Studio agent sessions in Postgres and answer follow-ups with history in context.\",\n  \"examples/models/lmstudio/image-agent\": \"Send image bytes to a llama3.2-vision model in LM Studio and stream the description.\",\n  \"examples/models/lmstudio/knowledge\": \"Query a PgVector knowledge base of PDF recipes from an agent running on LM Studio.\",\n  \"examples/models/lmstudio/structured-output\": \"Produce a MovieScript Pydantic object from an LM Studio model with output_schema.\",\n  \"examples/models/lmstudio/tool-use\": \"Add web search tools to an LM Studio agent and run sync and streaming queries.\",\n  \"examples/models/meta/llama-openai/basic\": \"Run Llama 4 Maverick through Meta's OpenAI-compatible API with sync, async, and streaming calls.\",\n  \"examples/models/meta/llama-openai/image-input-bytes\": \"Send image bytes to Llama 4 Maverick and fetch related news with web search tools.\",\n  \"examples/models/meta/llama-openai/image-input-file\": \"Attach a local image file to Llama 4 Maverick via the OpenAI-compatible client and stream a description.\",\n  \"examples/models/meta/llama-openai/knowledge\": \"Query a PgVector knowledge base of PDF recipes with Llama 4 Maverick over the OpenAI-compatible API.\",\n  \"examples/models/meta/llama-openai/metrics\": \"Stream a Llama 4 Maverick run with YFinance tools and inspect message, run, and session metrics.\",\n  \"examples/models/meta/llama-openai/storage\": \"Store sessions in a named Postgres table so Llama 4 Maverick keeps history across runs.\",\n  \"examples/models/meta/llama-openai/structured-output\": \"Return a MovieScript Pydantic model from Llama 4 Maverick through the OpenAI-compatible client.\",\n  \"examples/models/meta/llama-openai/tool-use\": \"Call YFinance tools from Llama 4 Maverick over the OpenAI-compatible API in sync and async modes.\",\n  \"examples/models/meta/llama/async-knowledge\": \"Query a PgVector knowledge base asynchronously with a Llama 4 Maverick agent.\",\n  \"examples/models/meta/llama/basic\": \"Run Llama 4 Maverick through Meta's Llama API with sync, async, and streaming calls.\",\n  \"examples/models/meta/llama/db\": \"Add a Postgres database to a Llama 4 Maverick agent to keep multi-turn chat history.\",\n  \"examples/models/meta/llama/image-input-bytes\": \"Pass a downloaded image as bytes to Llama 4 Maverick and search the web for related news.\",\n  \"examples/models/meta/llama/image-input-file\": \"Attach a local image file to a Llama 4 Maverick agent and stream its description.\",\n  \"examples/models/meta/llama/knowledge\": \"Answer recipe questions from a PgVector knowledge base with a Llama 4 Maverick agent.\",\n  \"examples/models/meta/llama/metrics\": \"Inspect per-message and run-level metrics from a Llama 4 Maverick run with YFinance tools.\",\n  \"examples/models/meta/llama/structured-output\": \"Generate a MovieScript Pydantic object from Llama 4 Maverick with a JSON schema output.\",\n  \"examples/models/meta/llama/tool-use\": \"Fetch stock prices with YFinance tools from a Llama 4 Maverick agent in sync and async modes.\",\n  \"examples/models/minimax/basic\": \"Run a MiniMax M3 agent across sync, async, and streaming response modes.\",\n  \"examples/models/minimax/structured-output\": \"Return a Pydantic MovieScript from MiniMax using JSON mode instead of native response_format.\",\n  \"examples/models/minimax/tool-use\": \"Equip a MiniMax M3 agent with WebSearchTools and stream a news query.\",\n  \"examples/models/mistral/basic\": \"Run a Mistral Small agent with plain and streaming print_response calls.\",\n  \"examples/models/mistral/image-bytes-input-agent\": \"Analyze raw image bytes with a Ministral 14B agent.\",\n  \"examples/models/mistral/image-compare-agent\": \"Compare two image URLs with a Ministral 14B agent.\",\n  \"examples/models/mistral/image-file-input-agent\": \"Analyze sample.jpeg with Ministral 14B and search for related news.\",\n  \"examples/models/mistral/image-ocr-with-structured-output\": \"Extract receipt data into a structured schema with Ministral 14B.\",\n  \"examples/models/mistral/mistral-small\": \"Query Mistral Small with WebSearchTools and stream a news summary.\",\n  \"examples/models/mistral/structured-output\": \"Produce a structured MovieScript from Mistral Large in both sync and async runs.\",\n  \"examples/models/mistral/structured-output-with-tool-use\": \"Combine web search with a Person output schema on Mistral Medium in a researcher agent.\",\n  \"examples/models/moonshot/basic\": \"Run a Moonshot Kimi K2 thinking agent with and without response streaming.\",\n  \"examples/models/moonshot/tool-use\": \"Add WebSearchTools to a Moonshot Kimi K2 agent and stream a current-events answer.\",\n  \"examples/models/n1n/basic\": \"Run GPT-4o through the N1N gateway, with and without streaming.\",\n  \"examples/models/n1n/tool-use\": \"Call web search tools from GPT-5 mini running through the N1N gateway.\",\n  \"examples/models/nebius/basic\": \"Prompt a Nebius agent synchronously, asynchronously, and with streaming.\",\n  \"examples/models/nebius/db\": \"Store Nebius agent sessions in Postgres and carry history across a follow-up question.\",\n  \"examples/models/nebius/knowledge\": \"Answer recipe questions with a Nebius Qwen3 agent over a PgVector knowledge base.\",\n  \"examples/models/nebius/structured-output\": \"Produce a MovieScript Pydantic object from Nebius Qwen3 with output_schema.\",\n  \"examples/models/nebius/tool-use\": \"Use WebSearchTools with Nebius Qwen3 across sync, async, and streaming runs.\",\n  \"examples/models/neosantara/basic\": \"Run a Grok model on Neosantara with sync, async, and streaming responses.\",\n  \"examples/models/neosantara/structured-output\": \"Generate a MovieScript with Neosantara Grok using output_schema and JSON mode.\",\n  \"examples/models/neosantara/tool-use\": \"Answer stock price questions on Neosantara with WebSearchTools, sync and async.\",\n  \"examples/models/nexus/basic\": \"Run Claude Sonnet 4 through the Nexus gateway, sync and async, with streaming.\",\n  \"examples/models/nexus/tool-use\": \"Call WebSearchTools from Claude Sonnet 4 routed through the Nexus gateway.\",\n  \"examples/models/nvidia/basic\": \"Run Llama 3.3 70B on the NVIDIA API in sync, async, and streaming modes.\",\n  \"examples/models/nvidia/tool-use\": \"Use web search tools with Llama 3.3 70B served through the NVIDIA API.\",\n  \"examples/models/ollama/chat/basic\": \"Run a local Llama 3.1 agent on Ollama in sync, async, and streaming modes.\",\n  \"examples/models/ollama/chat/db\": \"Persist Ollama agent sessions in Postgres and reuse history in a follow-up turn.\",\n  \"examples/models/ollama/chat/demo-deepseek-r1\": \"Ask a local DeepSeek-R1 model to write Python for quadratic equations and explain its reasoning.\",\n  \"examples/models/ollama/chat/demo-gemma\": \"Stream a short story about a local image from Gemma 3 12B running on Ollama.\",\n  \"examples/models/ollama/chat/demo-phi4\": \"Run Microsoft Phi-4 locally through Ollama with a one-shot story prompt.\",\n  \"examples/models/ollama/chat/demo-qwen\": \"Build an NVDA stock report with Qwen3 on Ollama using YFinanceTools.\",\n  \"examples/models/ollama/chat/image-agent\": \"Describe a local image file with Llama 3.2 Vision running on Ollama.\",\n  \"examples/models/ollama/chat/knowledge\": \"Search a PgVector knowledge base embedded with OllamaEmbedder from a local Llama agent.\",\n  \"examples/models/ollama/chat/reasoning-agent\": \"Enable reasoning on a local gpt-oss 120B agent and display the thinking steps.\",\n  \"examples/models/ollama/chat/set-client\": \"Configure the Ollama model with a custom ollama.Client instance.\",\n  \"examples/models/ollama/chat/set-temperature\": \"Set sampling temperature on an Ollama model through the options dict.\",\n  \"examples/models/ollama/chat/structured-output\": \"Return a MovieScript Pydantic object from a local Llama 3.2 model on Ollama.\",\n  \"examples/models/ollama/chat/tool-use\": \"Run web searches from a local Llama 3.2 agent with WebSearchTools on Ollama.\",\n  \"examples/models/openai/chat/agent-flex-tier\": \"Run o4-mini on the flex service tier by setting service_tier on OpenAIChat.\",\n  \"examples/models/openai/chat/audio-input-agent\": \"Send a WAV file fetched from a URL to gpt-audio and stream a text answer about it.\",\n  \"examples/models/openai/chat/audio-input-and-output-multi-turn\": \"Send audio input to gpt-audio and carry the conversation across turns with audio replies.\",\n  \"examples/models/openai/chat/audio-input-local-file-upload\": \"Upload a local MP3 file to gpt-audio and get a text description of the audio.\",\n  \"examples/models/openai/chat/audio-output-agent\": \"Get text and audio output from gpt-audio and save each spoken reply to a WAV file.\",\n  \"examples/models/openai/chat/audio-output-stream\": \"Stream pcm16 audio from gpt-audio, printing the transcript while writing frames to a WAV file.\",\n  \"examples/models/openai/chat/basic\": \"Run a gpt-4o agent in sync, async, and streaming modes with OpenAIChat.\",\n  \"examples/models/openai/chat/basic-stream-metrics\": \"Stream a response, then read run-level and per-message metrics from the last run output.\",\n  \"examples/models/openai/chat/db\": \"Store session history in Postgres so a follow-up question can reference the previous answer.\",\n  \"examples/models/openai/chat/generate-images\": \"Legacy DalleTools example that generates an image and reads its URL from the run output.\",\n  \"examples/models/openai/chat/image-agent\": \"Analyze an image from a URL and search the web for related news, streaming the reply.\",\n  \"examples/models/openai/chat/image-agent-bytes\": \"Pass an image as raw bytes and combine vision with web search for current news.\",\n  \"examples/models/openai/chat/image-agent-file\": \"Attach a local image by filepath, with auto-detected or explicit MIME types.\",\n  \"examples/models/openai/chat/image-agent-with-memory\": \"Analyze an image with web search, keeping chat history so follow-ups can reference it.\",\n  \"examples/models/openai/chat/knowledge\": \"Load a PDF into PgVector-backed knowledge and answer questions from its contents.\",\n  \"examples/models/openai/chat/metrics\": \"Print per-message and aggregated run metrics after a YFinance tool call.\",\n  \"examples/models/openai/chat/pdf-input-bytes\": \"Attach a PDF as raw bytes and have gpt-5-mini summarize its contents.\",\n  \"examples/models/openai/chat/pdf-input-local\": \"Attach a local PDF by filepath and query its contents with gpt-4o.\",\n  \"examples/models/openai/chat/pdf-input-url\": \"Attach a PDF by URL and ask the agent for a recipe from the file.\",\n  \"examples/models/openai/chat/reasoning-o3-mini\": \"Run o3-mini with high reasoning effort and YFinance tools to write a stock report.\",\n  \"examples/models/openai/chat/structured-output\": \"Compare JSON mode, strict, and guided structured output for a Pydantic movie schema.\",\n  \"examples/models/openai/chat/tool-use\": \"Answer questions with web search tools in sync, streaming, and async modes.\",\n  \"examples/models/openai/chat/verbosity-control\": \"Control response length with the gpt-5 verbosity parameter in a finance report agent.\",\n  \"examples/models/openai/chat/with-retries\": \"Configure model retries with delays and exponential backoff against a failing model ID.\",\n  \"examples/models/openai/responses/agent-flex-tier\": \"Run o4-mini on the flex service tier with the OpenAIResponses model.\",\n  \"examples/models/openai/responses/basic\": \"Run a gpt-4o agent through the Responses API in sync, async, and streaming modes.\",\n  \"examples/models/openai/responses/db\": \"Persist Responses API sessions in Postgres and answer a history-dependent follow-up.\",\n  \"examples/models/openai/responses/deep-research-agent\": \"Run OpenAI's o4-mini-deep-research model to produce a cited research report.\",\n  \"examples/models/openai/responses/file-input-direct\": \"Attach files by URL, local path, or raw bytes to a Responses API agent.\",\n  \"examples/models/openai/responses/image-agent\": \"Describe an image from a URL and fetch related news via web search with OpenAIResponses.\",\n  \"examples/models/openai/responses/image-agent-bytes\": \"Pass image bytes to a Responses API agent that pairs vision with web search.\",\n  \"examples/models/openai/responses/image-agent-file\": \"Attach a local image by filepath to a Responses API agent, overriding the MIME type if needed.\",\n  \"examples/models/openai/responses/image-agent-with-memory\": \"Reference an earlier image in follow-up questions by adding history to context.\",\n  \"examples/models/openai/responses/knowledge\": \"Answer questions from a PDF stored in PgVector knowledge with OpenAIResponses.\",\n  \"examples/models/openai/responses/pdf-input-local\": \"Query a local PDF with gpt-5.2 using the built-in file_search tool across two turns.\",\n  \"examples/models/openai/responses/pdf-input-url\": \"Summarize a PDF by URL with file_search and web search, then read citations from the stored session.\",\n  \"examples/models/openai/responses/reasoning-o3-mini\": \"Stream a stock report from o3-mini with YFinance tools via the Responses API.\",\n  \"examples/models/openai/responses/structured-output\": \"Generate a MovieScript Pydantic model via JSON mode, strict, and guided structured output.\",\n  \"examples/models/openai/responses/structured-output-with-tools\": \"Combine web search tools with a Pydantic output schema on gpt-5-mini.\",\n  \"examples/models/openai/responses/tool-use\": \"Answer a news question with GPT-4o on the Responses API using WebSearchTools.\",\n  \"examples/models/openai/responses/tool-use-gpt-5\": \"Fetch a live stock price with gpt-5 and cached YFinance tool results.\",\n  \"examples/models/openai/responses/tool-use-o3\": \"Call YFinance tools from o3 through the OpenAI Responses API to fetch a live stock price.\",\n  \"examples/models/openai/responses/verbosity-control\": \"Raise GPT-5 verbosity to high and generate a detailed stock comparison with YFinance tools.\",\n  \"examples/models/openai/responses/websearch-builtin-tool\": \"Combine the built-in web_search_preview tool with FileTools to search and save results to disk.\",\n  \"examples/models/openrouter/chat/basic\": \"Run a minimal OpenRouter agent with sync, streaming, and async response variants.\",\n  \"examples/models/perplexity/basic\": \"Run a basic Perplexity sonar-pro agent and print sync, streaming, and async responses.\",\n  \"examples/models/perplexity/knowledge\": \"Answer questions from a PDF knowledge base in PgVector using a Perplexity sonar-pro agent.\",\n  \"examples/models/perplexity/structured-output\": \"Return a typed MovieScript from Perplexity sonar-pro with a Pydantic output schema.\",\n  \"examples/models/perplexity/web-search\": \"Use Perplexity sonar-pro's native web search to answer a current news question.\",\n  \"examples/models/portkey/basic\": \"Route a basic agent through the Portkey AI gateway with sync, streaming, and async runs.\",\n  \"examples/models/portkey/structured-output\": \"Get a typed MovieScript response through Portkey using a Pydantic output schema.\",\n  \"examples/models/portkey/tool-use\": \"Call web search tools from an agent routed through the Portkey gateway.\",\n  \"examples/models/requesty/basic\": \"Run a minimal agent through the Requesty gateway with sync, streaming, and async calls.\",\n  \"examples/models/requesty/structured-output\": \"Return a typed MovieScript through Requesty with a Pydantic output schema.\",\n  \"examples/models/requesty/tool-use\": \"Stream a web-search agent through the Requesty gateway, sync and async.\",\n  \"examples/models/sambanova/basic\": \"Run Llama 3.1 8B on SambaNova with sync, streaming, and async response calls.\",\n  \"examples/models/siliconflow/basic\": \"Run gpt-oss-120b on SiliconFlow with sync, streaming, and async response calls.\",\n  \"examples/models/siliconflow/structured-output\": \"Use JSON mode on SiliconFlow to parse a Pydantic MovieScript from gpt-oss-120b.\",\n  \"examples/models/siliconflow/tool-use\": \"Run a SiliconFlow web-search agent with tool calls and debug output visible.\",\n  \"examples/models/together/basic\": \"Run a current Together chat model with sync, streaming, and async calls.\",\n  \"examples/models/together/image-agent\": \"Describe an image from a URL with a current Together vision model.\",\n  \"examples/models/together/image-agent-bytes\": \"Send local image bytes to a current Together vision model and stream the description.\",\n  \"examples/models/together/image-agent-with-memory\": \"Keep image context across turns with a current Together vision model and agent history.\",\n  \"examples/models/together/reasoning-agent\": \"Select a current Together reasoning model and print its reasoning steps.\",\n  \"examples/models/together/structured-output\": \"Return a Pydantic MovieScript through Together with a model that supports structured outputs.\",\n  \"examples/models/together/tool-use\": \"Give a function-calling Together model web search tools and run it sync, streaming, and async.\",\n  \"examples/models/vercel/basic\": \"Run a minimal agent on Vercel's v0 model with sync, streaming, and async calls.\",\n  \"examples/models/vercel/image-agent\": \"Pass an image URL to Vercel's v0 model and search the web for related news.\",\n  \"examples/models/vercel/knowledge\": \"Query a PgVector knowledge base built from a PDF with an agent on Vercel's v0 model.\",\n  \"examples/models/vercel/tool-use\": \"Stream a web-search agent on Vercel's v0 model, sync and async.\",\n  \"examples/models/vertexai/claude/basic\": \"Run Claude Sonnet 4 on Vertex AI with sync, streaming, and async calls.\",\n  \"examples/models/vertexai/claude/basic-with-timeout\": \"Set a five-second request timeout on Claude Sonnet 4 running on Vertex AI.\",\n  \"examples/models/vertexai/claude/db\": \"Persist session history in SQLite so a Vertex AI Claude agent can answer follow-ups.\",\n  \"examples/models/vertexai/claude/image-input-bytes\": \"Send a downloaded image as bytes to Claude on Vertex AI and search the web about it.\",\n  \"examples/models/vertexai/claude/image-input-url\": \"Pass an image URL to Claude on Vertex AI and search the web for more context.\",\n  \"examples/models/vertexai/claude/knowledge\": \"Answer questions from a PgVector knowledge base with Claude on Vertex AI.\",\n  \"examples/models/vertexai/claude/pdf-input-bytes\": \"Summarize a PDF passed as raw bytes to Claude Sonnet 4 on Vertex AI.\",\n  \"examples/models/vertexai/claude/pdf-input-local\": \"Summarize a local PDF file passed by path to Claude Sonnet 4 on Vertex AI.\",\n  \"examples/models/vertexai/claude/structured-output\": \"Return a typed MovieScript from Claude on Vertex AI with a Pydantic output schema.\",\n  \"examples/models/vertexai/claude/thinking\": \"Enable extended thinking with a 1024-token budget on Claude Sonnet 4 via Vertex AI.\",\n  \"examples/models/vertexai/claude/tool-use\": \"Give Claude on Vertex AI web search tools and run it sync, streaming, and async.\",\n  \"examples/models/vllm/basic\": \"Run a Qwen2.5-7B agent on a local vLLM server in sync, async, and streaming modes.\",\n  \"examples/models/vllm/db\": \"Store a vLLM agent's session history in Postgres and reuse it across turns.\",\n  \"examples/models/vllm/structured-output\": \"Return a typed MovieScript object from a vLLM agent with a Pydantic output schema.\",\n  \"examples/models/vllm/tool-use\": \"Add web search tools to a vLLM agent and stream responses sync and async.\",\n  \"examples/models/xai/basic\": \"Run a Grok agent in sync, async, and streaming modes with xAI.\",\n  \"examples/models/xai/finance-agent\": \"Use YFinanceTools with Grok to report stock prices, analyst recommendations, company data, and news.\",\n  \"examples/models/xai/image-agent\": \"Describe an image URL with Grok 4.5 and search for related news.\",\n  \"examples/models/xai/image-agent-bytes\": \"Analyze downloaded image bytes with Grok 4.5.\",\n  \"examples/models/xai/image-agent-with-memory\": \"Analyze an image with Grok 4.5 and retain history for follow-up questions.\",\n  \"examples/models/xai/live-search-agent\": \"Enable xAI Live Search on Grok with search parameters and citations for a news digest.\",\n  \"examples/models/xai/live-search-agent-stream\": \"Stream a world news digest from Grok using xAI Live Search with citations.\",\n  \"examples/models/xai/reasoning-agent\": \"Combine ReasoningTools and YFinance on Grok 4.3 to write a stock report with visible reasoning.\",\n  \"examples/models/xai/structured-output\": \"Get a typed MovieScript object back from Grok with a Pydantic output schema.\",\n  \"examples/models/xai/tool-use\": \"Build a web search agent on Grok and run it sync, streaming, and async.\",\n  \"examples/reasoning/agents/is-9-11-bigger-than-9-9\": \"Compare plain, built-in, and DeepSeek-backed reasoning agents on the 9.11 vs 9.9 test.\",\n  \"examples/reasoning/models/anthropic/async-reasoning-stream\": \"Stream Claude extended-thinking reasoning events asynchronously with stream_events.\",\n  \"examples/reasoning/models/anthropic/basic-reasoning\": \"Compare a plain Claude agent with one using extended thinking, then inspect reasoning_content.\",\n  \"examples/reasoning/models/anthropic/basic-reasoning-stream\": \"Stream reasoning events from Claude with extended thinking enabled.\",\n  \"examples/reasoning/models/azure-ai-foundry/reasoning-model-deepseek\": \"Use DeepSeek-R1 on Azure AI Foundry as the reasoning model behind a GPT-4o agent.\",\n  \"examples/reasoning/models/azure-ai-foundry/reasoning-model-stream-deepseek\": \"Stream reasoning events from a DeepSeek-R1 reasoning model on Azure AI Foundry.\",\n  \"examples/reasoning/models/azure-openai/basic-reasoning-stream\": \"Stream reasoning deltas from an Azure OpenAI GPT-4.1 reasoning model while the default chat model writes the answer.\",\n  \"examples/reasoning/models/azure-openai/o3-mini-with-tools\": \"Pair o3-mini on Azure OpenAI with YFinance tools to write a stock comparison report.\",\n  \"examples/reasoning/models/azure-openai/reasoning-model-gpt-4-1\": \"Delegate reasoning to GPT-4.1 while GPT-4o-mini writes the answer, both on Azure OpenAI.\",\n  \"examples/reasoning/models/deepseek/ethical-dilemma\": \"Use deepseek-reasoner to think through an ethical dilemma before GPT-4o responds.\",\n  \"examples/reasoning/models/deepseek/plan-itinerary\": \"Plan a Los Angeles to Las Vegas itinerary with DeepSeek reasoning feeding a GPT-4o agent.\",\n  \"examples/reasoning/models/gemini/async-reasoning-stream\": \"Stream Gemini thinking asynchronously with thinking_budget and include_thoughts set.\",\n  \"examples/reasoning/models/gemini/basic-reasoning\": \"Compare Gemini 2.5 Flash with thinking disabled and a 1,024-token budget, then inspect reasoning content.\",\n  \"examples/reasoning/models/gemini/basic-reasoning-stream\": \"Stream reasoning events from Gemini 2.5 Flash with a thinking budget enabled.\",\n  \"examples/reasoning/models/groq/deepseek-plus-claude\": \"Route reasoning through Qwen3-32B on Groq while Claude writes the final answer.\",\n  \"examples/reasoning/models/groq/fast-reasoning\": \"Time Llama 3.3 70B on Groq answering a math problem and check its reasoning depth.\",\n  \"examples/reasoning/models/groq/or-9-9\": \"Groq reasoning model works through the classic 9.11 vs 9.9 comparison.\",\n  \"examples/reasoning/models/ollama/local-reasoning\": \"Run QwQ and DeepSeek-R1 reasoning models locally through Ollama on a math task.\",\n  \"examples/reasoning/models/ollama/reasoning-model-deepseek\": \"Pair a local Llama 3.2 agent with DeepSeek-R1 on Ollama as its reasoning model.\",\n  \"examples/reasoning/models/openai/o3-mini\": \"Stream o3-mini working through the trolley problem with an ASCII diagram.\",\n  \"examples/reasoning/models/openai/o3-mini-with-tools\": \"Give o3-mini web search tools to research and write a stock comparison report.\",\n  \"examples/reasoning/models/openai/reasoning-effort\": \"Set reasoning_effort to high on o3-mini for a tool-assisted stock report.\",\n  \"examples/reasoning/models/openai/reasoning-model-gpt-4-1\": \"Use GPT-4.1 as the reasoning model behind a GPT-4o-mini agent via the Responses API.\",\n  \"examples/reasoning/models/openai/reasoning-stream\": \"Stream reasoning events from o3-mini on a history essay with stream_events enabled.\",\n  \"examples/reasoning/models/openai/reasoning-summary\": \"Request an automatic reasoning summary from o4-mini while it writes a stock report.\",\n  \"examples/reasoning/models/vertex-ai/basic-reasoning-stream\": \"Stream Gemini 2.5 Flash reasoning on Vertex AI with a fixed 1,024-token thinking budget and thought summaries.\",\n  \"examples/reasoning/models/xai/reasoning-effort\": \"Raise reasoning_effort on grok-3-mini-fast for a YFinance-backed stock report.\",\n  \"examples/reasoning/teams/finance-team-chain-of-thought\": \"Team leader with reasoning=True runs web and finance agents on a tariff impact analysis.\",\n  \"examples/reasoning/teams/knowledge-tool-team\": \"Give a team leader KnowledgeTools over a LanceDB knowledge base of a Paul Graham essay.\",\n  \"examples/reasoning/teams/reasoning-finance-team\": \"Claude team leader uses ReasoningTools to coordinate web and finance agents on tariff analysis.\",\n  \"examples/reasoning/tools/azure-openai-reasoning-tools\": \"Azure OpenAI agent pairs ReasoningTools with web search to write an NVDA vs TSLA report.\",\n  \"examples/reasoning/tools/capture-reasoning-content-knowledge-tools\": \"Capture reasoning_content from KnowledgeTools runs in streaming and non-streaming modes.\",\n  \"examples/reasoning/tools/capture-reasoning-content-reasoning-tools\": \"Inspect reasoning_content on a non-streaming RunOutput and the final streaming event.\",\n  \"examples/reasoning/tools/cerebras-llama-reasoning-tools\": \"Use ReasoningTools with a Cerebras agent to work through the fox, chicken, and grain puzzle.\",\n  \"examples/reasoning/tools/claude-reasoning-tools\": \"Claude Sonnet combines ReasoningTools and web search for a semiconductor market analysis.\",\n  \"examples/reasoning/tools/gemini-finance-agent\": \"Gemini agent uses ReasoningTools and YFinanceTools to compare NVDA and TSLA in a report.\",\n  \"examples/reasoning/tools/gemini-reasoning-tools\": \"Gemini 2.5 Pro runs think and analyze steps with YFinanceTools for an NVDA vs TSLA report.\",\n  \"examples/reasoning/tools/groq-llama-finance-agent\": \"Llama 4 Scout on Groq uses the think tool as a scratchpad while writing a finance report.\",\n  \"examples/reasoning/tools/ibm-watsonx-reasoning-tools\": \"IBM WatsonX Llama agent applies ReasoningTools to a river-crossing logic puzzle.\",\n  \"examples/reasoning/tools/knowledge-tools\": \"Agent thinks, searches, and analyzes over Agno docs stored in a LanceDB knowledge base.\",\n  \"examples/reasoning/tools/llama-reasoning-tools\": \"Llama 4 Maverick pairs ReasoningTools with YFinanceTools to report on the NVDA stock price.\",\n  \"examples/reasoning/tools/memory-tools\": \"Trip-planner agent stores and recalls user memories in SQLite with MemoryTools.\",\n  \"examples/reasoning/tools/ollama-reasoning-tools\": \"Local Llama 3.2 via Ollama uses ReasoningTools and web search for an NVDA vs TSLA report.\",\n  \"examples/reasoning/tools/openai-reasoning-tools\": \"GPT-4o combines few-shot ReasoningTools with web search to compare NVDA and TSLA.\",\n  \"examples/reasoning/tools/reasoning-tools\": \"GPT-4o agent uses the think and analyze tools to solve a river-crossing logic puzzle.\",\n  \"examples/reasoning/tools/vercel-reasoning-tools\": \"Vercel v0 model pairs ReasoningTools with web search to write a TSLA report.\",\n  \"examples/reasoning/tools/workflow-tools\": \"Agent drives a blog-post workflow through WorkflowTools with think and analyze steps.\",\n  \"examples/storage/dynamodb/dynamo-for-agent\": \"Store agent sessions and runs in DynamoDB using AWS credentials from the environment.\",\n  \"examples/storage/in-memory/in-memory-storage-for-agent\": \"Keep agent sessions in an in-memory database that clears when the process exits.\",\n  \"examples/storage/in-memory/in-memory-storage-for-team\": \"HackerNews research team stores sessions in an in-memory database with a structured Article output.\",\n  \"examples/storage/mongo/async-mongo/async-mongodb-for-agent\": \"Persist agent sessions in MongoDB with AsyncMongoDb and async run methods.\",\n  \"examples/storage/mongo/async-mongo/async-mongodb-for-team\": \"HackerNews team persists sessions in MongoDB with AsyncMongoDb and aprint_response.\",\n  \"examples/storage/mongo/async-mongo/async-mongodb-for-workflow\": \"Two-step content workflow stores its sessions in MongoDB through AsyncMongoDb.\",\n  \"examples/storage/mysql/mysql-for-agent\": \"Store agent sessions in MySQL and add history to context across runs.\",\n  \"examples/storage/mysql/mysql-for-team\": \"HackerNews research team stores its sessions in MySQL with structured Article output.\",\n  \"examples/storage/postgres/async-postgres/async-postgres-for-agent\": \"Persist agent sessions in Postgres with AsyncPostgresDb and async runs.\",\n  \"examples/storage/postgres/async-postgres/async-postgres-for-team\": \"HackerNews team persists sessions in Postgres using AsyncPostgresDb.\",\n  \"examples/storage/postgres/postgres-for-agent\": \"Persist agent sessions in Postgres and carry conversation history across runs.\",\n  \"examples/storage/postgres/postgres-for-team\": \"HackerNews research team stores sessions in Postgres with a structured Article output.\",\n  \"examples/storage/singlestore/singlestore-for-agent\": \"Store agent sessions in SingleStore using connection settings from environment variables.\",\n  \"examples/storage/singlestore/singlestore-for-team\": \"HackerNews team persists its sessions in SingleStore configured from environment variables.\",\n  \"examples/storage/sqlite/async-sqlite/async-sqlite-for-agent\": \"Persist agent sessions in SQLite with AsyncSqliteDb and async run methods.\",\n  \"examples/storage/sqlite/async-sqlite/async-sqlite-for-team\": \"HackerNews team persists sessions in a SQLite file using AsyncSqliteDb.\",\n  \"examples/storage/sqlite/async-sqlite/async-sqlite-for-workflow\": \"Two-step content workflow stores its sessions in SQLite through AsyncSqliteDb.\",\n  \"examples/storage/sqlite/sqlite-for-agent\": \"Store agent sessions in a SQLite file and recall earlier messages in the conversation.\",\n  \"examples/storage/sqlite/sqlite-for-team\": \"HackerNews research team stores sessions in a SQLite file with structured Article output.\",\n  \"examples/storage/surrealdb/surrealdb-for-agent\": \"Claude agent persists sessions in SurrealDB over a WebSocket connection.\",\n  \"examples/storage/surrealdb/surrealdb-for-team\": \"Persist a HackerNews research team's sessions in SurrealDB with structured Article output.\",\n  \"examples/storage/surrealdb/surrealdb-for-workflow\": \"Persist workflow sessions in SurrealDB for a two-step research and content planning pipeline.\",\n  \"examples/teams/basics/caching\": \"Cache team leader and member responses in two layers.\",\n  \"examples/teams/modes/coordinate/basic\": \"Coordinate mode: the team leader delegates to a researcher and writer, then synthesizes the answer.\",\n  \"examples/teams/modes/tasks/basic\": \"Tasks mode: the leader splits a request into planner, writer, and editor tasks and runs them in order.\",\n  \"examples/teams/task-mode/basic-task-mode\": \"A tasks-mode team decomposes a briefing request into research, writing, and review tasks.\",\n  \"examples/tools/baidusearch-tools\": \"Search Baidu in English and Chinese and return the top three results on a topic.\",\n  \"examples/tools/bravesearch-tools\": \"Fetch the latest news on a topic with BraveSearchTools, enabling specific or all functions.\",\n  \"examples/tools/brightdata-tools\": \"Scrape a webpage as Markdown with BrightDataTools, using include and exclude tool filters.\",\n  \"examples/tools/calcom-tools\": \"Use Calcom scheduling with Agno Agents.\",\n  \"examples/tools/calculator-tools\": \"Solve step-by-step arithmetic with CalculatorTools, filtering functions via include and exclude lists.\",\n  \"examples/tools/discord-tools\": \"Send messages, read history, inspect channels, and delete messages with DiscordTools.\",\n  \"examples/tools/composio-tools\": \"Give an agent a Composio GitHub action and ask it to star a repository.\",\n  \"examples/tools/confluence-tools\": \"List Confluence spaces, read page content, and create new pages with ConfluenceTools.\",\n  \"examples/tools/custom-tools\": \"Pass plain Python functions returning dicts, generators, Pydantic models, and dataclasses as tools.\",\n  \"examples/tools/docker-tools\": \"Configure DockerTools with include_tools and exclude_tools for container-management, delete-free, and full-access agents.\",\n  \"examples/tools/docling-tools/basic-examples\": \"Convert PDFs, DOCX, HTML, and more to Markdown, JSON, YAML, and other formats with DoclingTools.\",\n  \"examples/tools/docling-tools/ocr-example\": \"Convert PDFs with DoclingTools using forced full-page EasyOCR in Portuguese and English.\",\n  \"examples/tools/docling-tools/paths\": \"Helper module that resolves test document paths for the Docling tool examples.\",\n  \"examples/tools/docling-tools/run\": \"Entry point that runs the basic conversion and OCR Docling examples in sequence.\",\n  \"examples/tools/duckdb-tools\": \"Query a remote IMDB movie CSV with SQL using DuckDbTools to compute average ratings.\",\n  \"examples/tools/duckduckgo-tools\": \"Toggle DuckDuckGo search and news functions, plus WebSearchTools for other backends like Yandex.\",\n  \"examples/tools/email-tools\": \"Send an email from an agent with EmailTools configured with sender credentials and a receiver.\",\n  \"examples/tools/exa-tools\": \"Search, answer, and find similar pages with ExaTools, restricting results to specific news domains.\",\n  \"examples/tools/exceptions/retry-tool-call\": \"Raise RetryAgentRun from a tool to make the model retry until the shopping list has three items.\",\n  \"examples/tools/exceptions/stop-agent-exception\": \"Raise StopAgentRun inside a tool to halt the agent run before it completes.\",\n  \"examples/tools/fal-tools\": \"Generate a video from a text prompt with FalTools and the fal-ai/hunyuan-video model.\",\n  \"examples/tools/giphy-tools\": \"Search Giphy for a fitting GIF with GiphyTools, limiting results and enabled functions.\",\n  \"examples/tools/gitlab-tools\": \"List and summarize GitLab merge requests and issues with read-only GitlabTools functions.\",\n  \"examples/tools/googlesheets-tools\": \"Read a configured range with GoogleSheetsTools using service-account or OAuth credentials.\",\n  \"examples/tools/hackernews-tools\": \"Pull top HackerNews stories with HackerNewsTools and stream a report on trending startups.\",\n  \"examples/tools/jinareader-tools\": \"Fetch a GitHub page with JinaReaderTools and have the agent summarize its content.\",\n  \"examples/tools/jira-tools\": \"Search Jira issues, fetch details, and log work with different JiraTools configurations.\",\n  \"examples/tools/knowledge-tool\": \"Let an agent and team write new facts into a PgVector knowledge base with update_knowledge.\",\n  \"examples/tools/linkup-tools\": \"Answer a current-events question with web results from LinkupTools.\",\n  \"examples/tools/lumalabs-tools\": \"Generate videos from text prompts or images with LumaLabTools and the Luma AI API.\",\n  \"examples/tools/mcp-tools\": \"Connect MCPTools to the filesystem MCP server over a stdio session and summarize a file.\",\n  \"examples/tools/mcp/agno-mcp\": \"Query the Agno docs MCP server over streamable HTTP with MCPTools and a Claude agent.\",\n  \"examples/tools/mcp/dynamic-headers/client\": \"Send per-user headers to an MCP server with a header_provider that reads RunContext fields.\",\n  \"examples/tools/mcp/dynamic-headers/server\": \"FastMCP server with a greet tool that reads user and tenant IDs from incoming HTTP headers.\",\n  \"examples/tools/mcp/include-exclude-tools\": \"Filter which MCP server tools an agent can use with include_tools and exclude_tools.\",\n  \"examples/tools/mcp/include-tools\": \"Limit an MCP filesystem server to three functions with the include_tools filter.\",\n  \"examples/tools/mcp/local-server/client\": \"Launch a local FastMCP server by command and connect an agent to it with MCPTools.\",\n  \"examples/tools/mcp/mcp-toolbox-demo/agent-os\": \"Serve a hotel booking assistant in AgentOS using MCPToolbox hotel and booking toolsets.\",\n  \"examples/tools/mcp/mcp-toolbox-demo/hotel-management-typesafe\": \"Typed hotel search with Pydantic input and output schemas over MCPToolbox database tools.\",\n  \"examples/tools/mcp/mcp-toolbox-demo/hotel-management-workflows\": \"Chain hotel search and booking agents in a workflow, each with its own scoped MCP Toolbox toolset.\",\n  \"examples/tools/mcp/oxylabs\": \"Connect a Gemini agent to the Oxylabs MCP server to scrape a careers page for job titles.\",\n  \"examples/tools/mcp/pipedream-linkedin\": \"Review a legacy LinkedIn MCP integration that uses Pipedream's retired per-app SSE endpoint.\",\n  \"examples/tools/mcp/qdrant\": \"Store and retrieve agent output in a Qdrant collection through the Qdrant MCP server.\",\n  \"examples/tools/models-lab-tools\": \"Configure ModelsLabTools agents for image, video, and audio generation, then run image and sound-effect prompts.\",\n  \"examples/tools/moviepy-video-tools\": \"Extract audio, transcribe it, and embed SRT captions into a video with MoviePyVideoTools.\",\n  \"examples/tools/newspaper-tools\": \"Read and summarize an article from a URL with NewspaperTools.\",\n  \"examples/tools/newspaper4k-tools\": \"Summarize a blog post from its URL with the Newspaper4kTools article reader.\",\n  \"examples/tools/openbb-tools\": \"Pull stock prices, company news, and price targets with OpenBBTools function toggles.\",\n  \"examples/tools/other/add-tool-after-initialization\": \"Attach a new tool to an existing agent at runtime with add_tool().\",\n  \"examples/tools/other/cache-tool-calls\": \"Cache YFinance tool results with cache_results=True to skip repeat API calls.\",\n  \"examples/tools/other/include-exclude-tools\": \"Filter toolkit functions with include_tools and exclude_tools on CalculatorTools and WebSearchTools.\",\n  \"examples/tools/other/include-exclude-tools-custom-toolkit\": \"Restrict a custom customer database Toolkit to read-only functions with include_tools.\",\n  \"examples/tools/other/stop-after-tool-call\": \"End the run after web_search executes with stop_after_tool_call_tools and show the raw result.\",\n  \"examples/tools/oxylabs-tools\": \"Search Google through Oxylabs and summarize the top results with OxylabsTools.\",\n  \"examples/tools/parallel/parallel-tools\": \"Answer research questions with the Parallel web search API and stream the response.\",\n  \"examples/tools/postgres-tools\": \"Inspect and query a Postgres database with PostgresTools, including read-only tool filtering.\",\n  \"examples/tools/pubmed-tools\": \"Search PubMed for medical research papers with PubmedTools function toggles.\",\n  \"examples/tools/python-function-as-tool\": \"Pass a plain Python function to an agent as a tool to fetch top HackerNews stories.\",\n  \"examples/tools/redshift-tools\": \"Use Amazon Redshift tools with Agno agents.\",\n  \"examples/tools/replicate-tools\": \"Generate images and video on Replicate models with the generate_media tool.\",\n  \"examples/tools/resend-tools\": \"Send an email from your agent with ResendTools and a configured from address.\",\n  \"examples/tools/searxng-tools\": \"Query a self-hosted SearxNG instance at localhost:53153 for web, news, and science results.\",\n  \"examples/tools/serpapi-tools\": \"Run Google and YouTube searches through SerpAPI with per-function enable flags.\",\n  \"examples/tools/shell-tools\": \"Let an agent run shell commands to list directory contents with ShellTools.\",\n  \"examples/tools/sleep-tools\": \"Pause agent execution for a set number of seconds with SleepTools.\",\n  \"examples/tools/sql-tools\": \"List tables and query a Postgres database from a connection URL with SQLTools.\",\n  \"examples/tools/tavily-tools\": \"Search the web and extract page content with Tavily, tuning depth and output format.\",\n  \"examples/tools/tool-calls-accesing-agent\": \"Read agent.dependencies inside a tool by accepting the agent as a parameter.\",\n  \"examples/tools/tool-decorator/async-tool-decorator\": \"Define an async generator tool with @tool and stream its yielded results.\",\n  \"examples/tools/tool-decorator/cache-tool-calls\": \"Combine cache_results and stop_after_tool_call on a @tool-decorated function.\",\n  \"examples/tools/tool-decorator/stop-after-tool-call\": \"Stop the agent run immediately after a tool call with @tool(stop_after_tool_call=True).\",\n  \"examples/tools/tool-decorator/tool-decorator\": \"Decorate generator functions and class static methods with @tool, including show_result and async variants.\",\n  \"examples/tools/tool-decorator/tool-decorator-on-class-method\": \"Apply @tool to Toolkit class methods so decorated functions stay bound to the instance.\",\n  \"examples/tools/tool-decorator/tool-decorator-with-instructions\": \"Attach usage instructions, a name, and a description to a tool via @tool parameters.\",\n  \"examples/tools/tool-hooks/pre-and-post-hooks\": \"Log tool arguments and results with pre_hook and post_hook callbacks, sync and async.\",\n  \"examples/tools/web-tools\": \"Expand a shortened URL and describe its destination with WebTools.\",\n  \"examples/tools/webbrowser-tools\": \"Search for a page with WebSearchTools and open it in a local browser with WebBrowserTools.\",\n  \"examples/tools/websearch-tools\": \"Search the web and news with WebSearchTools across DuckDuckGo, Google, Bing, and Brave backends.\",\n  \"examples/tools/wikipedia-tools\": \"Search Wikipedia and optionally add results to a knowledge base with WikipediaTools.\",\n  \"examples/tools/youtube-tools\": \"Fetch YouTube video captions and answer questions about the video with YouTubeTools.\",\n  \"examples/tools/zendesk-tools\": \"Answer support questions by searching Zendesk help center articles with ZendeskTools.\",\n  \"examples/workflows/advanced-concepts/file-propagation/file-generation-workflow\": \"Generate a PDF with FileGenerationTools and propagate it to the next workflow step for analysis.\",\n  \"examples/basics/agent-with-storage\": \"Add SQLite session storage so the Finance Agent remembers conversations across runs.\",\n  \"examples/basics/multi-agent-team\": \"Coordinate bull and bear analyst agents under a team leader that synthesizes a balanced investment recommendation.\",\n  \"examples/basics/sequential-workflow\": \"Chain data gathering, analysis, and report writing agents into a three-step stock research workflow.\",\n  \"examples/basics/run\": \"Start an AgentOS server that exposes every quickstart agent, team, and workflow at http://localhost:7777.\",\n  \"examples/agents/basics/basic-agent\": \"Create a minimal agent with an OpenAIResponses model and print a streamed response.\",\n  \"examples/agents/basics/agent-with-instructions\": \"Steer agent responses with an instructions string that enforces a three-bullet answer format.\",\n  \"examples/agents/basics/agent-with-tools\": \"Give an agent DuckDuckGo web search tools and stream a summary of a recent AI safety headline.\",\n  \"examples/agents/input-output/response-as-variable\": \"Capture agent.run() output in a RunOutput variable and inspect it with pprint.\",\n  \"examples/agents/dependencies/dependencies-in-context\": \"Resolve a HackerNews fetch function as a runtime dependency and add its output to the agent's context.\",\n  \"examples/agents/dependencies/dependencies-in-tools\": \"Access dependencies passed to Agent.run() from inside a tool through the injected run_context parameter.\",\n  \"examples/agents/tools/tool-choice\": \"Compare tool_choice values none, auto, and a forced get_weather call across three agents.\",\n  \"examples/agents/state-and-session/session-state-basic\": \"Maintain a shopping list in session state and update it from a tool through RunContext.\",\n  \"examples/agents/state-and-session/persistent-session\": \"Persist agent sessions in Postgres with a fixed session_id so history carries across runs.\",\n  \"examples/agents/state-and-session/session-state-events\": \"Stream an agent run and read the final session state from RunCompletedEvent.\",\n  \"examples/agents/state-and-session/session-state-manual-update\": \"Modify session state between runs with get_session_state() and update_session_state().\",\n  \"examples/agents/state-and-session/session-state-advanced\": \"Manage a shopping list in session_state with tools that add, remove, and list items through RunContext.\",\n  \"examples/agents/state-and-session/agentic-session-state\": \"Let the agent update its own session_state shopping list with enable_agentic_state.\",\n  \"examples/agents/state-and-session/session-options\": \"Use history in context during runs while store_history_messages=False keeps history messages out of the database.\",\n  \"examples/agents/state-and-session/session-summary\": \"Enable session summaries with enable_session_summaries or a custom SessionSummaryManager, stored in Postgres.\",\n  \"examples/agents/state-and-session/last-n-session-messages\": \"Search a user's previous sessions with search_past_sessions, capped at the last two by num_past_sessions_to_search.\",\n  \"examples/agents/knowledge/rag-custom-embeddings\": \"Build agentic RAG over multilingual documents with SentenceTransformer embeddings and a BAAI reranker in PgVector.\",\n  \"examples/agents/guardrails/openai-moderation\": \"Block flagged input with OpenAIModerationGuardrail as a pre-hook, including custom category lists and image moderation.\",\n  \"examples/agents/guardrails/pii-detection\": \"Block SSNs, credit cards, emails, and phone numbers with PIIDetectionGuardrail, or mask them with mask_pii=True.\",\n  \"examples/agents/guardrails/prompt-injection\": \"Block prompt injection and jailbreak attempts with PromptInjectionGuardrail as a pre-hook.\",\n  \"examples/agents/guardrails/mixed-hooks\": \"Combine a logging pre-hook with PIIDetectionGuardrail; blocked runs return RunStatus.error instead of raising.\",\n  \"examples/agents/human-in-the-loop/confirmation-required\": \"Gate a custom HackerNews tool behind requires_confirmation and resume the paused run with agent.continue_run().\",\n  \"examples/agents/multimodal/image-to-text\": \"Pass a local sample.jpg to GPT-5.2 with OpenAIResponses and print a three-sentence story about the image.\",\n  \"examples/agents/multimodal/image-to-structured-output\": \"Turn an image URL into a Pydantic MovieScript with GPT-5.2 and stream the parsed result.\",\n  \"examples/agents/multimodal/image-to-image\": \"Give an agent FalTools and a source image URL so it can generate a transformed image with Fal's image_to_image model.\",\n  \"examples/agents/multimodal/image-to-audio\": \"Write a story about a local image with gpt-4o, then narrate it with gpt-audio and save the WAV to tmp/.\",\n  \"examples/agents/multimodal/audio-input-output\": \"Send a WAV recording to gpt-audio and save the model's spoken reply to tmp/result.wav with write_audio_to_file().\",\n  \"examples/agents/multimodal/audio-sentiment-analysis\": \"Analyze speaker sentiment in a recorded conversation with Gemini, then ask a follow-up question answered from SQLite-backed session history.\",\n  \"examples/agents/multimodal/media-input-for-tool\": \"Declare a files parameter on a tool and Agno injects the files passed to the agent, with send_media_to_model=False keeping the PDF out of the model request.\",\n  \"examples/agents/multimodal/video-caption\": \"Extract a video's audio, transcribe it with OpenAITools, and embed the generated SRT captions back into the video with MoviePyVideoTools.\",\n  \"examples/agents/context-management/instructions-with-state\": \"Pass a callable as instructions to build the agent's system message from session_state on every run.\",\n  \"examples/agents/hooks/post-hook-output\": \"Validate agent responses for completeness, tone, safety, and length with post_hooks that raise OutputCheckError.\",\n  \"examples/agents/hooks/pre-hook-input\": \"Validate financial-advice input with a pre-hook, then detect blocked calls by checking for RunStatus.error.\",\n  \"examples/agents/hooks/session-state-hooks\": \"Track conversation topics in session_state by updating RunContext from a pre_hook.\",\n  \"examples/agents/hooks/stream-hook\": \"Send a notification from a post_hook after a streamed agent response completes.\",\n  \"examples/agents/advanced/background-execution\": \"Start a background agent run that returns PENDING immediately, then poll for completion or cancel it.\",\n  \"examples/agents/advanced/basic-agent-events\": \"Stream run lifecycle, tool call, and content events from an agent with stream_events=True.\",\n  \"examples/agents/advanced/compression-events\": \"Stream CompressionStarted and CompressionCompleted events from an agent using compress_tool_results=True.\",\n  \"examples/agents/advanced/custom-logging\": \"Configure the default agno.utils.log logger with a custom Python logger.\",\n  \"examples/agents/advanced/reasoning-agent-events\": \"Stream reasoning_started, reasoning_step, and reasoning_completed events from a reasoning-enabled agent run.\",\n  \"examples/agents/advanced/retries\": \"Configure retries, delay_between_retries, and exponential_backoff on an Agent.\",\n  \"examples/agents/advanced/tool-call-compression\": \"Set compress_tool_results=True to compress tool call results and save context on search-heavy runs.\",\n  \"examples/agents/skills/basic-skills\": \"Load skills from a local directory with LocalSkills and use them in a code review agent.\",\n  \"examples/agents/advanced/combined-metrics\": \"Inspect per-model metrics detail keys and session-level metrics for an agent using reasoning, compression, memory, culture, summary, and eval models.\",\n  \"examples/agents/fallback-models/mid-run-fallback\": \"Trigger a fallback when the primary model fails mid-run, after a tool call has already executed.\",\n  \"examples/agents/fallback-models/fallback-callback\": \"Register a FallbackConfig callback that fires when a fallback model serves the request.\",\n  \"examples/agents/checkpointing/crash-recovery\": \"Crash an in-flight run with SIGKILL, then resume it from the last persisted tool-batch checkpoint via /continue.\",\n  \"examples/agents/checkpointing/tool-error-persistence\": \"Persist an agent conversation through tool exceptions and model failures, then retry the failed run in place with `Agent.acontinue_run()`.\",\n  \"examples/agents/checkpointing/checkpoint-endpoints\": \"List a run's checkpoint boundaries and fetch truncated snapshots via the AgentOS checkpoint endpoints, then resume from a chosen index.\",\n  \"examples/agents/time-travel/fork-run\": \"Fork a completed run at a message boundary into a new sibling run in the same session, leaving the original intact.\",\n  \"examples/agents/fork-session/fork-session\": \"Fork a session into a new session with a fresh session_id and fresh run IDs while the source session stays untouched.\",\n  \"examples/teams/basics/respond-directly-router-team\": \"Route each question to the member agent that speaks its language with TeamMode.route and return that member's answer directly.\",\n  \"examples/teams/basics/delegate-to-all-members\": \"Broadcast the same research task to every member with TeamMode.broadcast, fanning it out to Reddit and HackerNews researchers.\",\n  \"examples/teams/basics/share-member-interactions\": \"Give support team members each other's requests and responses from the current run with share_member_interactions=True.\",\n  \"examples/teams/basics/concurrent-member-agents\": \"Stream member events with timestamps while the team leader delegates to HackerNews and news agents concurrently.\",\n  \"examples/teams/basics/history-of-members\": \"Give each member its own conversation history with add_history_to_context while the leader routes questions by language.\",\n  \"examples/teams/modes/tasks-stream\": \"Stream TaskCreated, TaskUpdated, TaskIteration, and TaskStateUpdated events to render a real-time task list.\",\n  \"examples/teams/human-in-the-loop/confirmation-required-async\": \"Pause an asynchronous team run for confirmation, approve its production-deployment tool call, and resume it with team.acontinue_run().\",\n  \"examples/teams/checkpointing/crash-recovery\": \"Simulate a hard crash during a team run, then resume its last RUNNING checkpoint with `Team.acontinue_run()`.\",\n  \"examples/teams/checkpointing/tool-error-persistence\": \"Persist a team conversation through tool exceptions and model failures, then retry the failed run in place with `Team.acontinue_run()`.\",\n  \"examples/learning/entity-memory/entity-relationships\": \"Create and query typed relationships between entities with AGENTIC entity memory.\",\n  \"examples/workflows/cel-expressions/router/cel-additional-data-route\": \"Uses additional_data.route to let the caller specify which step to run, useful when the routing decision is made upstream (e.g. UI).\",\n  \"examples/tools/google/gmail/daily-digest\": \"Summarize recent emails into a structured daily digest grouped by category and tagged with priority.\",\n  \"examples/workflows/cel-expressions/router/cel-using-step-choices\": \"Uses step_choices[0], step_choices[1], etc. to reference steps by their position in the choices list, rather than hardcoding step names.\",\n  \"examples/workflows/conditional-execution/condition-with-parallel\": \"Run Hacker News, web, and Exa research conditions in parallel before two synthesis steps.\",\n  \"examples/workflows/conditional-branching/nested-choices\": \"Route to a single step or a nested sequence of steps based on selector logic.\",\n  \"examples/workflows/conditional-branching/router-basic\": \"Route to HackerNews or web research based on topic keywords.\",\n  \"examples/workflows/conditional-branching/router-with-loop\": \"Route to simple web search or iterative deep research loops based on topic complexity.\",\n  \"examples/workflows/conditional-branching/selector-media-pipeline\": \"Route media requests to OpenAI image generation or Gemini video generation.\",\n  \"examples/workflows/conditional-branching/selector-types\": \"Use string, step, list, and nested return types in router selectors.\",\n  \"examples/workflows/conditional-branching/step-choices-parameter\": \"Access available choices in selector functions to dynamically route between steps.\",\n  \"examples/workflows/conditional-branching/string-selector\": \"Return a step name string from a router selector.\",\n  \"examples/workflows/advanced-concepts/background-execution/background-poll\": \"Run a workflow in async background mode and poll run status until completion.\",\n  \"examples/workflows/advanced-concepts/background-execution/websocket-client\": \"Build an interactive WebSocket client for authenticating, starting workflows, and rendering streamed workflow events.\",\n  \"examples/workflows/advanced-concepts/background-execution/websocket-server\": \"Run background workflows and stream workflow events over WebSocket.\",\n  \"examples/workflows/advanced-concepts/early-stopping/early-stop-basic\": \"Implement early termination with `StepOutput(stop=True)` across direct steps, `Steps` containers, and agent/function workflows.\",\n  \"examples/workflows/advanced-concepts/early-stopping/early-stop-condition\": \"Stop an entire workflow when a Condition branch detects policy violations.\",\n  \"examples/workflows/advanced-concepts/early-stopping/early-stop-loop\": \"Halt a Loop when a safety-check step detects sensitive AI-related content.\",\n  \"examples/workflows/advanced-concepts/early-stopping/early-stop-parallel\": \"Stop an entire workflow from a Parallel step when checking input for dangerous content.\",\n  \"examples/workflows/advanced-concepts/guardrails/prompt-injection\": \"Block prompt-injection attacks using PromptInjectionGuardrail in async workflows.\",\n  \"examples/workflows/advanced-concepts/history/continuous-execution\": \"Give a tutoring workflow access to its three most recent runs for a continuing conversation.\",\n  \"examples/workflows/advanced-concepts/history/history-in-function\": \"Access workflow history from function steps to analyze content strategy and prevent topic overlap.\",\n  \"examples/workflows/advanced-concepts/history/intent-routing-with-history\": \"Route customer requests to specialist steps while maintaining shared workflow history for context continuity.\",\n  \"examples/workflows/advanced-concepts/history/step-history\": \"Control workflow-level and step-level history access for multi-step conversation workflows.\",\n  \"examples/workflows/advanced-concepts/previous-step-outputs/access-previous-outputs\": \"Access output from multiple prior steps using both named steps and implicit step keys.\",\n  \"examples/workflows/advanced-concepts/run-control/cancel-run\": \"Cancel a workflow run from another thread while it's executing.\",\n  \"examples/workflows/advanced-concepts/run-control/deep-copy\": \"Create independent workflow copies with isolated state using `deep_copy()`.\",\n  \"examples/workflows/advanced-concepts/run-control/event-storage\": \"Store workflow events while filtering out high-volume event types.\",\n  \"examples/workflows/advanced-concepts/run-control/executor-events\": \"Suppress intermediate executor events while preserving terminal executor events when streaming.\",\n  \"examples/workflows/advanced-concepts/run-control/metrics\": \"Extract workflow and step-level execution metrics including duration and token counts.\",\n  \"examples/workflows/advanced-concepts/run-control/remote-workflow\": \"Execute workflows hosted on a remote AgentOS server using async calls.\",\n  \"examples/workflows/advanced-concepts/run-control/workflow-cli\": \"Use `Workflow.cli_app()` for interactive command-line workflow runs.\",\n  \"examples/workflows/advanced-concepts/run-control/workflow-serialization\": \"Persist workflows with `to_dict()`, `save()`, and `load()`.\",\n  \"examples/workflows/advanced-concepts/session-state/rename-session\": \"Auto-generate a workflow session name after a run.\",\n  \"examples/workflows/advanced-concepts/session-state/state-in-condition\": \"Use workflow session state in `Condition` evaluator and executor functions.\",\n  \"examples/workflows/advanced-concepts/session-state/state-in-function\": \"Read and mutate workflow session state in custom function executors.\",\n  \"examples/workflows/advanced-concepts/session-state/state-in-router\": \"Use router selectors with workflow session state for adaptive routing.\",\n  \"examples/workflows/advanced-concepts/session-state/state-with-agent\": \"Share workflow session state across agent tool calls.\",\n  \"examples/workflows/advanced-concepts/session-state/state-with-team\": \"Share session state across team and agent steps in project workflows.\",\n  \"examples/workflows/advanced-concepts/structured-io/image-input\": \"Pass images to workflows and chain vision analysis with web search.\",\n  \"examples/workflows/advanced-concepts/structured-io/input-schema\": \"Validate workflow input with Pydantic schemas.\",\n  \"examples/workflows/advanced-concepts/structured-io/pydantic-input\": \"Pass Pydantic model instances as workflow input.\",\n  \"examples/workflows/advanced-concepts/structured-io/structured-io-agent\": \"Chain agents with structured output schemas.\",\n  \"examples/workflows/advanced-concepts/structured-io/structured-io-function\": \"Return strings and BaseModel objects from custom function steps in workflows with structured I/O and data flow analysis.\",\n  \"examples/workflows/advanced-concepts/structured-io/structured-io-team\": \"Use structured output schemas across team steps in multi-step workflows.\",\n  \"examples/workflows/advanced-concepts/tools/workflow-tools\": \"Expose workflows as tools for agent execution.\",\n  \"examples/workflows/advanced-concepts/workflow-agent/basic-workflow-agent\": \"WorkflowAgent decides whether to execute steps or answer from history.\",\n  \"examples/workflows/advanced-concepts/workflow-agent/workflow-agent-with-condition\": \"Add conditional steps to a WorkflowAgent graph.\",\n  \"examples/workflows/advanced-concepts/nested-workflows/nested-workflow-auto-wrap\": \"Pass a Workflow directly in steps list for auto-wrapping without explicit Step() wrapper.\",\n  \"examples/workflows/advanced-concepts/nested-workflows/nested-workflow-with-condition\": \"Use Condition steps in nested workflows to conditionally branch execution based on content analysis.\",\n  \"examples/workflows/advanced-concepts/nested-workflows/nested-workflow-with-loop\": \"Use Loop steps in nested workflows to iteratively refine output until a quality threshold is met.\",\n  \"examples/workflows/advanced-concepts/nested-workflows/nested-workflow-with-router\": \"Use Router steps in nested workflows to delegate tasks to specialist agents based on input keywords.\",\n  \"examples/workflows/advanced-concepts/run-params/workflow-all-params\": \"Configure workflow behavior with metadata, dependencies, and session state in a content pipeline.\",\n  \"examples/workflows/advanced-concepts/run-params/workflow-dependencies\": \"Inject configuration through workflow dependencies to control agent behavior.\",\n  \"examples/workflows/advanced-concepts/session-state/job-application-tracker\": \"Extract job applications with structured output and persist them in session state across workflow runs.\",\n  \"examples/workflows/human-in-the-loop/confirmation/custom-function-step-confirmation\": \"Require confirmation for workflow steps with the @pause decorator.\",\n  \"examples/workflows/human-in-the-loop/decision-tree/decision-tree\": \"Build a multi-step decision tree with sequential Condition pauses at each branch.\",\n  \"examples/workflows/human-in-the-loop/decision-tree/multi-component-decision-tree\": \"Decision tree mixing Condition and Loop components with user confirmation at each pause point.\",\n  \"examples/workflows/human-in-the-loop/dual-level-hitl/step-confirmation-and-tool-confirmation\": \"Require step-level confirmation, then tool-level confirmation, on a single step.\",\n  \"examples/workflows/human-in-the-loop/dual-level-hitl/step-user-input-and-tool-confirmation\": \"Collect step-level user input, then require confirmation for the agent's tool call.\",\n  \"examples/workflows/human-in-the-loop/dual-level-hitl/condition-and-tool-confirmation\": \"Require confirmation for branch selection, then confirm the tool call inside the chosen branch.\",\n  \"examples/workflows/human-in-the-loop/dual-level-hitl/router-selection-and-tool-confirmation\": \"Router-level user selection of route, then tool confirmation within the chosen agent.\",\n  \"examples/workflows/human-in-the-loop/error/error-retry-skip\": \"Handle step errors by pausing workflow and letting user choose to retry or skip the step.\",\n  \"examples/workflows/human-in-the-loop/error/error-retry-skip-streaming\": \"Handle step errors with streaming event updates, pausing to let user retry or skip failed steps.\",\n  \"examples/workflows/human-in-the-loop/executor-hitl/agent-confirmation\": \"Pause workflow execution until the user confirms an agent tool call.\",\n  \"examples/workflows/human-in-the-loop/executor-hitl/agent-confirmation-stream\": \"Emit StepExecutorPausedEvent in workflow stream when agent tool needs confirmation.\",\n  \"examples/workflows/human-in-the-loop/executor-hitl/team-in-step\": \"Propagate a team member's tool-confirmation pause through the containing workflow.\",\n  \"examples/workflows/human-in-the-loop/executor-hitl/executor-continued-event\": \"The StepExecutorContinuedEvent: emitted when a paused executor (agent/team) resumes after executor-level HITL is resolved.\",\n  \"examples/workflows/human-in-the-loop/output-review/edit-output\": \"Allow humans to edit step output directly instead of retrying, avoiding extra LLM calls.\",\n  \"examples/workflows/human-in-the-loop/output-review/full-review-cycle\": \"Handle approve, reject with feedback and retry, and cancel decisions in one HITL workflow.\",\n  \"examples/workflows/human-in-the-loop/router/router-nested-choices\": \"Let users choose from pre-configured processing packages.\",\n  \"examples/workflows/human-in-the-loop/router/router-confirmation\": \"Automatically route requests and require user confirmation to proceed.\",\n  \"examples/workflows/human-in-the-loop/timeout/confirmation-timeout\": \"Apply automatic timeouts to pre-execution confirmation with HumanReview.\",\n  \"examples/workflows/human-in-the-loop/user-input/step-user-input-streaming\": \"Handle human-in-the-loop pauses with streaming workflow events.\",\n  \"examples/storage/dynamodb/dynamo-for-team\": \"Store team sessions and runs in DynamoDB with environment variables or constructor parameters.\",\n  \"examples/storage/examples/multi-user-multi-session\": \"Handle multiple users and sessions with SQLite-backed agent storage.\",\n  \"examples/storage/gcs/gcs-json-for-agent\": \"Store agent sessions in Google Cloud Storage as JSON blobs.\",\n  \"examples/storage/in-memory/in-memory-storage-for-workflow\": \"Use in-memory storage for Workflow demos.\",\n  \"examples/storage/mysql/async-mysql/async-mysql-for-agent\": \"Store agent sessions in async MySQL while using web search and conversation history.\",\n  \"examples/storage/mysql/async-mysql/async-mysql-for-team\": \"Store a HackerNews research team's structured article output in MySQL asynchronously.\",\n  \"examples/storage/mysql/async-mysql/async-mysql-for-workflow\": \"Store a multi-step research and blog-writing workflow in MySQL asynchronously.\",\n  \"examples/storage/postgres/postgres-for-workflow\": \"Store workflow sessions in PostgreSQL with a multi-step content creation example.\",\n  \"examples/storage/postgres/async-postgres/async-postgres-for-workflow\": \"Use async Postgres as the storage backend for a workflow.\",\n  \"examples/storage/redis/redis-for-agent\": \"Use Redis as the storage backend for an agent.\",\n  \"examples/storage/redis/redis-for-team\": \"Use Redis as the storage backend for a team.\",\n  \"examples/storage/redis/redis-for-workflow\": \"Use Redis as the storage backend for a workflow.\",\n  \"examples/storage/sqlite/sqlite-for-workflow\": \"Use SqliteDb as the session storage backend for a workflow.\",\n  \"examples/storage/persistent-session-storage\": \"Store and retrieve team sessions across runs using PostgreSQL.\",\n  \"examples/storage/session-summary\": \"Automatically summarize agent sessions to reduce context window usage.\",\n  \"examples/storage/chat-history\": \"Retrieve and inspect the full chat history from persisted agent sessions.\",\n  \"examples/knowledge/building-blocks/chunking-strategies\": \"Compare four chunking strategies for different content types and use cases.\",\n  \"examples/knowledge/building-blocks/hybrid-search\": \"Compare vector search, keyword search, and hybrid approaches to find the right strategy for your knowledge base.\",\n  \"examples/knowledge/building-blocks/reranking\": \"Implement two-stage retrieval with Cohere reranking to improve search result relevance.\",\n  \"examples/knowledge/production/error-handling\": \"Handle knowledge ingestion failures with skip_if_exists, batch error logging, and verification patterns.\",\n  \"examples/knowledge/advanced/custom-retriever\": \"Provide a custom retriever function for non-vector data sources such as SQL queries, API calls, or file lookups.\",\n  \"examples/knowledge/advanced/custom-chunking\": \"Implement a custom ParagraphChunking strategy that splits documents on double newlines for well-structured prose.\",\n  \"examples/knowledge/advanced/knowledge-tools\": \"Provide agents with think, search, and analyze tools for multi-step knowledge reasoning.\",\n  \"examples/knowledge/advanced/knowledge-protocol\": \"Implement KnowledgeProtocol for an in-memory source with custom context and synchronous or asynchronous document retrieval.\",\n  \"examples/knowledge/integrations/rag/agentic-rag-infinity-reranker\": \"Run hybrid LanceDB search with Cohere embeddings and a local Infinity reranker (BAAI/bge-reranker-base on localhost:7997) behind a Claude agent.\",\n  \"examples/knowledge/integrations/rag/agentic-rag-with-lightrag\": \"Ingest a PDF, a Wikipedia topic, and a URL into a LightRag-backed knowledge base and query it with an async agent.\",\n  \"examples/knowledge/integrations/readers/documents\": \"Load a local PDF, an Excel file via an explicit ExcelReader, and a remote PDF URL into Qdrant hybrid search, then query them with an agent.\",\n  \"examples/knowledge/integrations/readers/data\": \"Ingest inline CSV and JSON text with CSVReader and JSONReader into Qdrant hybrid search and query the rows with an agent.\",\n  \"examples/knowledge/integrations/readers/web\": \"Crawl docs.agno.com with WebsiteReader (max_depth=1, max_links=5) and auto-detect a remote PDF URL into Qdrant hybrid search.\",\n  \"examples/knowledge/integrations/readers/docling/docling-xlsx\": \"Read an XLSX spreadsheet with DoclingReader in HTML output mode and query products and prices through a LanceDB-backed agent.\",\n  \"examples/knowledge/integrations/vector-dbs/qdrant\": \"Run Qdrant-backed knowledge two ways: basic vector search, and hybrid search with a Cohere reranker.\",\n  \"examples/knowledge/integrations/vector-dbs/local\": \"Compare embedded ChromaDB and hybrid-search LanceDB knowledge bases over the same PDF, with graceful fallback when either package is missing.\",\n  \"examples/memory/agent-with-memory\": \"Persist user memories in PostgresDb with update_memory_on_run so the agent creates and updates them after each run.\",\n  \"examples/memory/agentic-memory\": \"Enable enable_agentic_memory so the agent itself creates, updates, and deletes user memories during a run.\",\n  \"examples/memory/memory-manager/custom-memory-instructions\": \"Run a custom OpenAI memory-capture scenario for academic interests, then a separate default Claude scenario over multi-turn messages.\",\n  \"examples/memory/multi-user-multi-session-chat-concurrent\": \"Run three user conversations concurrently with asyncio.gather while a shared agent persists user memories across sessions.\",\n  \"examples/learning/basics/b-entity-memory-agentic\": \"AGENTIC mode gives the agent search_entities, create_entity, add_fact, add_event, and add_relationship tools to track companies and contacts.\",\n  \"examples/learning/demo/agents\": \"Shared ops assistant configured with all six learning stores, backed by PostgresDb and a PgVector knowledge base.\",\n  \"examples/context/web-exa\": \"Run WebContextProvider with ExaBackend so an agent searches and fetches web pages through Exa's search + contents API.\",\n  \"examples/context/database-read-write\": \"Use one SQLite engine for read and write operations through DatabaseContextProvider.\",\n  \"examples/context/slack-search-media\": \"Run SlackContextProvider with Gemini sub-agents, enabling search_messages and opt-in media tools for file download and upload.\",\n  \"examples/context/multi-provider\": \"Compose filesystem, Exa MCP web, and SQLite database context providers on one agent, bracketing only the web provider's MCP session with asetup/aclose.\",\n  \"examples/context/engineering-briefing\": \"Synthesizes a briefing by chaining Slack, workspace, and Parallel web context providers, then posting the result back to Slack.\",\n  \"examples/context/wiki-filesystem\": \"Seeds a local markdown wiki and uses WikiContextProvider with FileSystemBackend so the agent writes and reads back a runbook with no auth or network.\",\n  \"examples/context/wiki-with-web\": \"Wires an ExaMCPBackend into WikiContextProvider so a single update_wiki call can search or fetch the web, digest the result, and file it as a wiki page.\",\n  \"examples/context/wiki-dual\": \"Mount two WikiContextProvider instances on one agent: a read-write company_knowledge wiki and a read-only company_voice wiki.\",\n  \"examples/context/gdrive-office\": \"Reads .docx, .xlsx and .pptx files from Google Drive via GoogleDriveContextProvider, extracting text with python-docx, openpyxl and python-pptx.\",\n  \"examples/context/wiki-agentos-streaming\": \"Serve a WikiContextProvider-backed agent on AgentOS at localhost:7777 so sub-agent tool calls and content stream live into os.agno.com.\",\n  \"examples/context/multi-context-streaming\": \"Attach two WikiContextProviders (architecture and operations) to one AgentOS agent and stream nested events from parallel sub-agent tool calls.\",\n  \"examples/models/openai/chat/access-memories-in-memory-completed-event\": \"Stream an agent run with stream_events=True and read user memories off the MemoryUpdateCompleted event, backed by PostgresDb.\",\n  \"examples/models/openai/chat/custom-role-map\": \"Point OpenAIChat at Mistral's base_url and remap the `model` role to `assistant` via role_map.\",\n  \"examples/models/openai/chat/memory\": \"Store user memories and session summaries in Postgres and print them after each turn.\",\n  \"examples/models/openai/chat/pdf-input-file-upload\": \"Attach a local PDF file to an OpenAI chat agent and ask it to suggest a recipe from the document.\",\n  \"examples/models/openai/chat/retry\": \"Set retries, delay_between_retries, and exponential_backoff on OpenAIChat using an invalid model ID to trigger them.\",\n  \"examples/models/openai/chat/text-to-speech-agent\": \"Generate speech with the OpenAITools speech toolkit and save the returned audio to tmp/speech_output.mp3.\",\n  \"examples/models/openai/responses/image-generation-agent\": \"Generate and save an image with OpenAITools and GPT Image 2.\",\n  \"examples/models/openai/responses/memory\": \"Store user memories and session summaries in Postgres while chatting with gpt-4o over the Responses API.\",\n  \"examples/models/openai/responses/zdr-reasoning-agent\": \"Run o4-mini with store=False and reasoning_summary=\\\"auto\\\", keeping multi-turn context in an InMemoryDb for Zero Data Retention.\",\n  \"examples/models/anthropic/betas\": \"Enable the 1M-token context beta on Claude via the betas parameter and print every beta available in the installed anthropic SDK.\",\n  \"examples/models/anthropic/image-input-file-upload\": \"Upload a PNG to the Anthropic Files API and pass the file handle to Claude as an image input.\",\n  \"examples/models/anthropic/image-input-local-file\": \"Download a PNG locally and pass it to Claude by filepath as an image input.\",\n  \"examples/models/anthropic/memory\": \"Store Claude agent memories and session summaries in Postgres across multiple turns.\",\n  \"examples/models/anthropic/pdf-input-file-upload\": \"Upload a PDF through the Anthropic Files API beta and pass the file handle to a Claude agent for summarization.\",\n  \"examples/models/anthropic/prompt-caching\": \"Enable Claude cache_system_prompt for a large system message and print cache write and read token counts across two runs.\",\n  \"examples/models/anthropic/prompt-caching-extended\": \"Set the extended-cache-ttl beta and extended_cache_time on Claude to hold the cached system prompt for one hour, and compare cache write and read tokens.\",\n  \"examples/models/anthropic/retry\": \"Configure Claude with retries, delay_between_retries, and exponential_backoff, using an invalid model ID to force the retry path.\",\n  \"examples/models/anthropic/structured-output-strict-tools\": \"Combine a strict-mode Function schema with a Pydantic output_schema so Claude validates both the tool input and the final response.\",\n  \"examples/models/anthropic/adaptive-thinking\": \"Let Claude 4.6 decide its own thinking depth with thinking={\\\"type\\\": \\\"adaptive\\\"} and an effort level.\",\n  \"examples/models/anthropic/append-trailing-user-message\": \"Append a trailing user turn with append_trailing_user_message when a Claude 4.6 conversation ends on an assistant message.\",\n  \"examples/models/anthropic/pydantic-tool-input\": \"Pass nested, Optional, Union, List, and deeply nested Pydantic models as tool inputs to Claude.\",\n  \"examples/models/google/gemini/agent-with-thinking-budget\": \"Cap Gemini 2.5 Pro's reasoning with thinking_budget and surface thought summaries with include_thoughts.\",\n  \"examples/models/google/gemini/external-url-input\": \"Pass a public HTTPS PDF URL straight to Gemini 3.5 Flash with File(url=...), no download step.\",\n  \"examples/models/google/gemini/file-upload-with-cache\": \"Upload a transcript with the Gemini Files API, cache it with a 5-minute TTL, and query the cached content to cut prompt tokens.\",\n  \"examples/models/google/gemini/gcs-file-input\": \"Summarize a PDF straight from a Google Cloud Storage gs:// URI using Gemini on Vertex AI, with no download or re-upload.\",\n  \"examples/models/google/gemini/imagen-tool\": \"Generate an image with the GeminiTools Imagen toolkit driven by a GPT-4o agent and save the result as a PNG.\",\n  \"examples/models/google/gemini/imagen-tool-advanced\": \"Generate an image with the Imagen 4 model through GeminiTools on Vertex AI and save the result as a PNG.\",\n  \"examples/models/google/gemini/pdf-input-file-upload\": \"Upload a PDF to the Gemini Files API, then ask an agent to summarize it and suggest a recipe from it.\",\n  \"examples/models/google/gemini/retry\": \"Configure retry count, retry delay, and exponential backoff on the Gemini model.\",\n  \"examples/models/google/gemini/s3-url-file-input\": \"Generate an S3 pre-signed URL with boto3 and pass it straight to Gemini to summarize the PDF without downloading it.\",\n  \"examples/models/google/gemini/vertexai\": \"Authenticate Gemini against Vertex AI with environment variables or explicit `project_id` and `location` parameters on the Gemini model.\",\n  \"examples/models/google/gemini-interactions/antigravity-environment-config\": \"Reuse an existing Antigravity sandbox by ID, or pass a full EnvironmentConfig to set sources and network rules for a new one.\",\n  \"examples/models/google/gemini-interactions/audio-understanding\": \"Send an audio clip URL to a GeminiInteractions agent and have it describe what it hears.\",\n  \"examples/models/google/gemini-interactions/basic\": \"Run a GeminiInteractions agent through sync, streaming, and async response calls.\",\n  \"examples/models/google/gemini-interactions/tool-use\": \"Give a GeminiInteractions agent WebSearchTools for live searches across sync, streaming, and async calls.\",\n  \"examples/models/mistral/image-transcribe-document-agent\": \"Transcribe a document image with a Ministral 14B agent.\",\n  \"examples/models/mistral/memory\": \"Persist user memories and session summaries for a Mistral Large agent in Postgres.\",\n  \"examples/models/mistral/retry\": \"Configure retry count, delay, and exponential backoff on a Mistral model.\",\n  \"examples/models/together/retry\": \"Configure retry count, delay, and exponential backoff on a Together model.\",\n  \"examples/models/fireworks/retry\": \"Inspect Fireworks retry settings in a pinned example whose invalid-model request returns a terminal error.\",\n  \"examples/models/aws/retry\": \"Configure retry logic with exponential backoff for AWS Bedrock models.\",\n  \"examples/models/aws/claude/adaptive-thinking\": \"Configure Claude 4.6 adaptive thinking with effort levels on AWS Bedrock.\",\n  \"examples/models/aws/claude/append-trailing-user-message\": \"Append trailing user messages when Claude 4.6+ ends with an assistant response.\",\n  \"examples/models/azure/retry\": \"Configure retry strategy, delay, and exponential backoff with AzureAIFoundry.\",\n  \"examples/models/vertexai/claude/betas\": \"Enable Anthropic beta features like context-management with Claude on Vertex AI.\",\n  \"examples/models/vertexai/claude/memory\": \"Store personalized memories and session summaries with Claude on Vertex AI using PostgreSQL.\",\n  \"examples/models/vertexai/claude/prompt-caching\": \"Cache the system prompt with Claude on Vertex AI to reduce processing time and costs.\",\n  \"examples/models/vertexai/retry\": \"Configure retry policy with exponential backoff on Vertex AI Claude models.\",\n  \"examples/models/vertexai/claude/adaptive-thinking\": \"Use adaptive thinking with effort levels to control reasoning depth on Claude VertexAI.\",\n  \"examples/models/ibm/retry\": \"Configure retry policy with exponential backoff on IBM WatsonX models.\",\n  \"examples/models/nvidia/retry\": \"Configure retries, retry delay, and exponential backoff on the NVIDIA model.\",\n  \"examples/models/cerebras/retry\": \"Retry failed Cerebras calls with exponential backoff by pointing an agent at an invalid model ID.\",\n  \"examples/models/xai/retry\": \"Retry a deliberately invalid Grok model request three times with a one-second initial delay and exponential backoff.\",\n  \"examples/models/ollama/chat/memory\": \"Persist user memories and session summaries in Postgres with a local qwen2.5 Ollama agent.\",\n  \"examples/models/ollama/chat/ollama-cloud\": \"Run gpt-oss:120b on Ollama Cloud with OLLAMA_API_KEY instead of a local Ollama server.\",\n  \"examples/models/ollama/chat/retry\": \"Set up retries with Ollama using the retries, delay_between_retries, and exponential_backoff parameters.\",\n  \"examples/models/ollama/responses/basic\": \"Use Ollama's OpenAI-compatible /v1/responses endpoint with an Agent.\",\n  \"examples/models/vllm/retry\": \"Retry a failing vLLM request three times with exponential backoff.\",\n  \"examples/models/lmstudio/memory\": \"Store and retrieve personalized user memories and conversation summaries in an LM Studio agent.\",\n  \"examples/models/lmstudio/retry\": \"Configure retry behavior with retries, delays, and exponential backoff in LM Studio models.\",\n  \"examples/models/llama-cpp/retry\": \"Configure LlamaCpp with retries, delay, and exponential backoff, using a wrong model ID to trigger them.\",\n  \"examples/models/aimlapi/retry\": \"Configure retry logic with exponential backoff for AIMLAPI requests.\",\n  \"examples/models/cloudflare/basic\": \"Run a Workers AI model through Cloudflare AI Gateway in sync, async, and streaming modes.\",\n  \"examples/models/cloudflare/switch-model\": \"Switch Cloudflare AI Gateway models by passing a catalog binding to `Cloudflare(id=...)` or a `cloudflare:` model string.\",\n  \"examples/models/cloudflare/tool-use\": \"Give a Cloudflare AI Gateway agent web search with WebSearchTools and stream the tool-calling run.\",\n  \"examples/models/cometapi/multi-model\": \"Test and compare different models available through CometAPI.\",\n  \"examples/models/cometapi/retry\": \"Set up retries with exponential backoff for CometAPI requests.\",\n  \"examples/models/dashscope/knowledge-tools\": \"Use OpenAI embeddings with LanceDB for hybrid search to build a knowledge-augmented DashScope agent.\",\n  \"examples/models/dashscope/retry\": \"Configure exponential backoff retries with delay for DashScope model requests.\",\n  \"examples/models/deepinfra/retry\": \"Set up retries with exponential backoff for DeepInfra model requests.\",\n  \"examples/models/internlm/retry\": \"Set up retries with InternLM model requests.\",\n  \"examples/models/langdb/retry\": \"Retry failed LangDB requests three times with exponential backoff using a deliberately invalid model ID.\",\n  \"examples/models/litellm/reasoning-agent\": \"Stream DeepSeek R1 reasoning_content through LiteLLM while the agent compares 9.11 and 9.9.\",\n  \"examples/models/litellm/retry\": \"Configure LiteLLM retries with exponential backoff using a deliberately invalid model ID.\",\n  \"examples/models/litellm/append-trailing-user-message\": \"Append a trailing user turn so Claude 4.6+ models that reject assistant prefill work through LiteLLM.\",\n  \"examples/models/litellm-openai/audio-input-agent\": \"Send an MP3 to gpt-audio through a local LiteLLM proxy with LiteLLMOpenAI.\",\n  \"examples/models/nebius/retry\": \"Trigger Nebius request retries with an invalid model id, using 3 retries and exponential backoff.\",\n  \"examples/models/nexus/retry\": \"Trigger Nexus retries with an invalid model ID using retries, delay_between_retries, and exponential_backoff.\",\n  \"examples/models/openrouter/chat/dynamic-model-router\": \"Configure an OpenRouter agent with an ordered fallback model list so a request failing on rate limits, timeouts, or overload retries on the next model.\",\n  \"examples/models/openrouter/chat/retry\": \"Configure OpenRouter with 3 retries, a 1s delay, and exponential backoff using a deliberately wrong model ID.\",\n  \"examples/models/perplexity/memory\": \"Persist Perplexity sonar-pro user memories and session summaries in Postgres across multiple runs.\",\n  \"examples/models/perplexity/retry\": \"Trigger Perplexity retries with exponential backoff by using a deliberately invalid model id.\",\n  \"examples/models/portkey/retry\": \"Configure Portkey with 3 retries, a 1-second delay, and exponential backoff.\",\n  \"examples/models/requesty/retry\": \"Trigger Requesty retries with a bad model id using 3 retries and exponential backoff.\",\n  \"examples/models/sambanova/retry\": \"Trigger SambaNova retries with an invalid model ID using 3 retries and exponential backoff.\",\n  \"examples/models/siliconflow/retry\": \"Trigger three exponential-backoff retries by pointing a SiliconFlow agent at an invalid model ID.\",\n  \"examples/models/vercel/retry\": \"Trigger V0 retries with a deliberately wrong model id, using retries, delay_between_retries, and exponential_backoff.\",\n  \"examples/models/xiaomi/structured-output\": \"Return a MovieScript Pydantic object from MiMo with `use_json_mode=True`, since MiMo supports JSON mode but not native json_schema structured outputs.\",\n  \"examples/tools/tool-hooks/tool-hook-in-toolkit\": \"Block deletion and retrieval of customer ID 123 with a validation tool_hook wrapped around a custom Toolkit.\",\n  \"examples/tools/tool-hooks/tool-hook-in-toolkit-with-state\": \"Swap a customer ID for the full profile stored in session_state by rewriting tool arguments inside a tool_hook.\",\n  \"examples/tools/tool-hooks/tool-hook-in-toolkit-with-state-nested\": \"Chain two nested tool hooks so the first swaps a customer ID for the profile stored in session_state before the toolkit function runs.\",\n  \"examples/tools/tool-hooks/tool-hooks-in-toolkit-nested\": \"Nest validation and logging hooks around toolkit tools to block customer ID 123 and strip the name field from results, in sync and async variants.\",\n  \"examples/tools/tool-decorator/tool-decorator-with-hook\": \"Attach a custom duration-logging hook to a tool via @tool(tool_hooks=[...]) to time each tool call.\",\n  \"examples/tools/tool-decorator/toolkit-per-tool-instructions\": \"Verify per-tool instructions from a bare @tool function and a Toolkit both reach agent._tool_instructions via parse_tools.\",\n  \"examples/tools/mcp/mcp-toolbox-demo/agent\": \"Load MCPToolbox hotel-management and booking-system toolsets, or load them manually with auth token getters and bound params, then drive an interactive CLI hotel agent.\",\n  \"examples/tools/mcp/sse-transport/client\": \"Connect an OpenAI agent to an SSE MCP server with MCPTools, and to SSE plus stdio servers together with MultiMCPTools.\",\n  \"examples/tools/mcp/airbnb\": \"Search Airbnb listings with an OpenAI gpt-4o agent connected to the @openbnb/mcp-server-airbnb stdio MCP server.\",\n  \"examples/tools/mcp/mcp-toolbox-for-db\": \"Connect an agent to an MCP Toolbox for Databases server and load hotel-management and booking-system toolsets via MCPToolbox.\",\n  \"examples/tools/mcp/pipedream-auth\": \"Call an authenticated Pipedream MCP server over streamable HTTP, passing a bearer token plus project and environment headers on behalf of an end user.\",\n  \"examples/tools/mcp/stagehand\": \"Drive a Browserbase Stagehand MCP browser to scrape Hacker News headlines and comments into a structured reader's digest.\",\n  \"examples/tools/mcp/bgpt\": \"Query the hosted BGPT MCP server over Streamable HTTP to search papers and surface methods, sample sizes, limitations, and conflicts of interest.\",\n  \"examples/tools/scrapegraph-tools\": \"Extract structured data, markdown, and raw HTML from a page using ScrapeGraphTools smartscraper, markdownify, and scrape.\",\n  \"examples/tools/scheduler-tools\": \"Create and manage schedule records from a standalone agent; this page covers record operations only.\",\n  \"examples/tools/trafilatura-tools\": \"Extract clean page text with TrafilaturaTools across txt, markdown, JSON and XML output, precision/recall tuning, metadata-only mode, crawling and HTML-to-text.\",\n  \"examples/tools/youcom-tools\": \"Search the web with YouTools, including a domain-allowlisted agent limited to cnbc.com, reuters.com and bloomberg.com.\",\n  \"examples/tools/arxiv-tools\": \"Configure ArxivTools with enable_search_arxiv / enable_read_arxiv_papers and all=True to control which ArXiv functions an agent can call.\",\n  \"examples/tools/file-generation-tools\": \"Generate JSON, CSV, PDF, TXT, DOCX and HTML files from agent responses with FileGenerationTools, persisting them to an output directory.\",\n  \"examples/tools/crawl4ai-tools\": \"Compare three Crawl4aiTools agents: default crawling with pruning, an explicit enable_crawl agent, and a raw full-page agent without pruning.\",\n  \"examples/tools/custom-tool-events\": \"Yield a custom CustomEvent subclass from an async @tool and consume it while streaming agent.arun().\",\n  \"examples/tools/async-generator-tool-with-pydantic-args\": \"Sync-generator, async-generator, and async-coroutine toolkit tools that each receive a Pydantic SearchParams model and stream custom progress events.\",\n  \"examples/tools/duckduckgo-tools-advanced\": \"Configure DuckDuckGoTools with timelimit, region, backend, fixed_max_results, and timeout to build week-scoped, region-localized, and news-only search agents.\",\n  \"examples/tools/llms-txt-tools\": \"Read an llms.txt index with LLMsTxtTools, then fetch only the documentation pages the agent decides are relevant.\",\n  \"examples/tools/llms-txt-tools-knowledge\": \"Bulk-load every page linked from an llms.txt index into a PgVector knowledge base, then answer questions with semantic search.\",\n  \"examples/tools/local-file-system-tools\": \"Write and read files in a target directory with LocalFileSystemTools, including a write-only agent that disables read_file.\",\n  \"examples/tools/perplexity-tools\": \"Search the web with PerplexitySearch, including a variant filtered by recency window and allowed domains.\",\n  \"examples/tools/scavio-tools\": \"Gate ScavioTools providers with enable_* flags to build web-only, commerce-only, and all-provider search agents over Google, YouTube, Amazon, Walmart, Reddit, TikTok, and Instagram.\",\n  \"examples/tools/searchapi-tools\": \"Configure SearchApiTools for Google web, News, Images, and YouTube SERP results using per-engine enable flags or all=True.\",\n  \"examples/tools/sofya-tools\": \"Toggle SofyaTools between web search, markdown URL extraction, and cited deep-research report generation.\",\n  \"examples/tools/twelvelabs-tools\": \"Answer questions about a video URL with TwelveLabs Pegasus analyze_video, and generate Marengo multimodal embeddings with embed_text and embed_video.\",\n  \"examples/tools/websearch-tools-advanced\": \"Configure WebSearchTools time limits, regions, and DDGS backends for text and news search.\",\n  \"examples/tools/x402scan-mcp-tools\": \"Pay for 100+ paid APIs autonomously with USDC on Base via the x402scan MCP server, with a single agent and a researcher/analyst team sharing one wallet.\",\n  \"examples/tools/models/azure-openai-tools\": \"Legacy AzureOpenAITools example for image generation with retired DALL-E deployments.\",\n  \"examples/tools/models/gemini-image-generation\": \"Generate an image with GeminiTools, then save the returned image content to disk.\",\n  \"examples/tools/models/openai-tools\": \"Transcribe audio and generate an image with OpenAITools and GPT Image 2.\",\n  \"examples/tools/spotify-tools\": \"Search tracks and create or update Spotify playlists with SpotifyTools and a Claude-powered DJ agent.\",\n  \"examples/tools/other/complex-input-types\": \"Pass nested Pydantic models and enums as tool arguments so the agent fills validated UserProfile and Task schemas.\",\n  \"examples/tools/other/session-state-tool\": \"Read and write run_context.session_state inside a @tool to cache the last FAQ answer across runs, persisted with SqliteDb.\",\n  \"examples/tools/other/stop-after-tool-call-dual-inheritance\": \"Confirm stop_after_tool_call_tools flags still apply when a Toolkit subclass also inherits from a non-Toolkit base class.\",\n  \"examples/tools/other/stop-after-tool-call-in-toolkit\": \"Register Toolkit methods with stop_after_tool_call_tools so one tool halts the agent run while the others continue.\",\n  \"examples/tools/seltz-tools\": \"Give an agent SeltzTools for fast, clean web data retrieval, with retrieved results logged for debugging.\",\n  \"examples/tools/website-tools\": \"Scrape and summarize a web page with WebsiteTools, which registers its read_url function when no knowledge base is attached.\",\n  \"examples/tools/google/drive/folder-organizer\": \"Browses Drive folders with list_files and uploads or downloads local files with write tools enabled.\",\n  \"examples/tools/google/gmail/basic\": \"Configure Gmail agents with include_tools/exclude_tools for read-only, send-blocked, label-management, and full-access variants.\",\n  \"examples/tools/parallel/market-research\": \"Configure ParallelTools Task API with text and JSON output schemas to produce cited market reports and structured industry data.\",\n  \"examples/tools/parallel/news-search\": \"Build general, tech, and finance search agents with ParallelTools using include_domains filters and result-count limits.\",\n  \"examples/tools/parallel/output-schemas\": \"Compare the Parallel Task API's auto, JSON, string, and text output schemas on one research query.\",\n  \"examples/tools/workspace-tools/workspace-tools-with-confirmation\": \"Approve or reject Workspace write/edit/delete tool calls from the console using pause/resume with a SQLite-backed agent.\",\n  \"examples/agent-os/demo\": \"Serve an AgentOS with a docs-MCP agent over PgVector knowledge plus a Postgres-backed web research team.\",\n  \"examples/agent-os/advanced-demo/demo\": \"Serve an AgentOS app combining the Sage and Agno Assist agents with the finance reasoning team, plus an AccuracyEval setup.\",\n  \"examples/agent-os/advanced-demo/mcp-demo\": \"Define a FastAPI lifespan that connects MCPTools to the GitHub MCP server over stdio.\",\n  \"examples/agent-os/advanced-demo/reasoning-model\": \"Stream a Claude thinking-model chain of thought live from an AgentOS reasoning agent.\",\n  \"examples/agent-os/background-tasks/background-hooks-decorator\": \"Opt individual hooks into background execution with @hook(run_in_background=True), leaving other post-hooks in the normal flow.\",\n  \"examples/agent-os/background-tasks/background-hooks-example\": \"Enable AgentOS-wide background hooks with run_hooks_in_background=True so all pre- and post-hooks run after the API response is sent.\",\n  \"examples/agent-os/background-tasks/background-hooks-team\": \"Attach a non-blocking post-hook to a researcher/writer Team that logs run id and content length after the response is sent.\",\n  \"examples/agent-os/background-tasks/background-hooks-workflow\": \"Run per-step agent post-hooks in the background across an analyzer/summarizer Workflow served by AgentOS.\",\n  \"examples/agent-os/background-tasks/evals-demo\": \"Register an AccuracyEval for a calculator agent backed by PostgresDb and browse eval runs through the AgentOS eval endpoints.\",\n  \"examples/agent-os/client/basic-client\": \"Fetch AgentOS config over HTTP with AgentOSClient and list registered agents, teams and workflows, then inspect one agent's model and tools.\",\n  \"examples/agent-os/client/knowledge-search\": \"Read the remote knowledge config, list indexed content items, and run a top-5 semantic search with scores via AgentOSClient.\",\n  \"examples/agent-os/client/memory-operations\": \"Full CRUD over a remote user's memories with AgentOSClient: create, list, get, update topics, read memory topics and per-user stats, then delete and verify.\",\n  \"examples/agent-os/client/run-evals\": \"Trigger accuracy, performance and reliability evals (with expected tool calls and argument matching) against a remote agent, then list eval runs and fetch one run's details.\",\n  \"examples/agent-os/client/session-management\": \"Create, list, inspect, rename, and delete AgentOS sessions and read their runs with AgentOSClient.\",\n  \"examples/agent-os/customize/custom-fastapi-app\": \"Mount AgentOS on an existing FastAPI base_app so a custom /customers route is served alongside AgentOS routes.\",\n  \"examples/agent-os/customize/custom-lifespan\": \"Pass an asynccontextmanager lifespan to AgentOS to run startup and shutdown hooks around the FastAPI app.\",\n  \"examples/agent-os/customize/handle-custom-events\": \"Yield a CustomEvent subclass from an async tool that reads session_state, streamed through AgentOS agent and team run endpoints.\",\n  \"examples/agent-os/customize/override-routes\": \"Use on_route_conflict=\\\"preserve_base_app\\\" so custom / and /health routes win over the conflicting AgentOS routes.\",\n  \"examples/agent-os/customize/pass-dependencies-to-agent\": \"Send a dependencies payload on an AgentOS run request to fill the {robot_name} placeholder in the agent's instructions.\",\n  \"examples/agent-os/dbs/agentos-default-db\": \"Share one PostgresDb across AgentOS and its agent as the default database, with PgVector knowledge and the Agno MCP server.\",\n  \"examples/agent-os/dbs/dynamo\": \"Back an AgentOS agent, team, and AccuracyEval with DynamoDb for sessions, memories, and summaries.\",\n  \"examples/agent-os/dbs/firestore\": \"Back an AgentOS agent, team, and AccuracyEval with FirestoreDb using explicit session, memory, eval, metrics, and knowledge collections.\",\n  \"examples/agent-os/dbs/gcs-json\": \"Back an AgentOS agent, team, and AccuracyEval with GcsJsonDb, storing all state as JSON objects in a GCS bucket.\",\n  \"examples/agent-os/dbs/mongo\": \"Run AgentOS on MongoDB with parallel MongoDb and AsyncMongoDb backends, each wired to an agent, team, and accuracy eval.\",\n  \"examples/agent-os/dbs/mysql\": \"Run AgentOS on MySQL with MySQLDb (pymysql) and AsyncMySQLDb (asyncmy) backends and custom session, eval, memory, and metrics tables.\",\n  \"examples/agent-os/dbs/postgres\": \"Run AgentOS on Postgres with PostgresDb and AsyncPostgresDb backends, the async setup adding a persisted workflow alongside the agent and team.\",\n  \"examples/agent-os/dbs/singlestore\": \"Back an AgentOS agent, team, and AccuracyEval with SingleStoreDb using custom session, eval, memory, and metrics tables.\",\n  \"examples/agent-os/dbs/sqlite\": \"Run an AgentOS agent, team, and AccuracyEval on a local SqliteDb file with named session, eval, memory, and metrics tables.\",\n  \"examples/agent-os/dbs/supabase\": \"Point PostgresDb at a Supabase project by building its connection URL from SUPABASE_PROJECT and SUPABASE_PASSWORD env vars.\",\n  \"examples/agent-os/dbs/surreal\": \"Store AgentOS sessions and knowledge contents in SurrealDb while embedding knowledge into a PgVector table.\",\n  \"examples/agent-os/integrations/shopify-demo\": \"AgentOS sales-analyst agent that queries Shopify orders, products and customers via ShopifyTools, with chat quick prompts for top sellers and revenue trends.\",\n  \"examples/agent-os/interfaces/slack/agent-with-user-memory\": \"Slack personal assistant that captures names, hobbies, and preferences with MemoryManager and recalls them in later chats.\",\n  \"examples/agent-os/interfaces/slack/hitl-audit-flow\": \"Incident-commander agent forced to call a tool every turn, with approval, external-execution, and user-input pauses in Slack.\",\n  \"examples/agent-os/interfaces/slack/hitl-confirmation\": \"Pause irreversible subscription cancellation for an Approve or Deny decision in Slack.\",\n  \"examples/agent-os/interfaces/slack/hitl-external-execution\": \"Slack agent whose kubectl tool is external_execution=True, so the engineer runs the command locally and pastes output back for analysis against internal runbooks.\",\n  \"examples/agent-os/interfaces/slack/hitl-required-approval\": \"Persist approval records for schema migrations so Slack approvals survive restarts and leave an audit trail.\",\n  \"examples/agent-os/interfaces/slack/hitl-user-feedback\": \"Travel concierge that calls ask_user via UserFeedbackTools, so Slack renders a single TaskCard of multi- and single-select preference questions whose answers feed the itinerary draft.\",\n  \"examples/agent-os/interfaces/slack/hitl-user-input\": \"Support-intake agent that drafts a ticket title and body but declares priority and component as user_input_fields, pausing Slack with an input form the requester fills before the tool runs.\",\n  \"examples/agent-os/interfaces/slack/team-hitl-confirmation\": \"Propagate a team member's weather-tool confirmation pause to Approve and Deny buttons in Slack.\",\n  \"examples/agent-os/interfaces/slack/team-hitl-external-execution-simple\": \"Slack team whose member agent has an external-execution send_email tool; the run pauses, Slack shows the tool name and arguments, and the user supplies the result.\",\n  \"examples/agent-os/interfaces/slack/team-hitl-team-tool-simple\": \"Pause a Slack team run for Approve/Deny when the team-level deployment tool is called.\",\n  \"examples/agent-os/interfaces/slack/team-hitl-user-input-simple\": \"Pause a Slack team run so the user fills in destination and budget fields before the trip-planning tool executes.\",\n  \"examples/agent-os/interfaces/agui/agentic-chat\": \"Combine an external-execution change_background tool with a get_weather tool that returns card-ready data.\",\n  \"examples/agent-os/interfaces/agui/backend-tool-rendering\": \"Agent whose get_weather tool returns structured city/temperature/humidity data for the AG-UI frontend to render as a weather card.\",\n  \"examples/agent-os/interfaces/agui/human-in-the-loop\": \"Agent whose generate_task_steps tool requires user confirmation, returning a toggleable step list for the AG-UI Dojo step selector.\",\n  \"examples/agent-os/interfaces/agui/shared-state\": \"Sync recipe session_state edits to an AG-UI frontend through STATE_DELTA events.\",\n  \"examples/agent-os/interfaces/agui/state-events\": \"Serves a recipe agent over AG-UI at /shared_state, where enable_agentic_state lets the model call update_session_state to sync recipe fields to the UI.\",\n  \"examples/agent-os/interfaces/agui/team-state-events\": \"Serves a recipe-creator plus nutrition-advisor Team over AG-UI at /shared_state, with enable_agentic_state syncing the shared recipe session state to the UI.\",\n  \"examples/agent-os/interfaces/agui/tool-based-generative-ui\": \"Haiku agent that calls an externally executed generate_haiku tool, passing Japanese and English lines plus an image name and CSS gradient for the frontend to render as a card.\",\n  \"examples/agent-os/interfaces/telegram/reasoning-agent\": \"Telegram bot that pairs ReasoningTools think/analyze steps with DuckDuckGo search, running on OpenAIChat gpt-5.2 with SQLite session history.\",\n  \"examples/agent-os/knowledge/agentos-knowledge\": \"Serve an AgentOS agent over hybrid-search PgVector knowledge, toggling between PostgresDb and AsyncPostgresDb with a USE_ASYNC flag.\",\n  \"examples/agent-os/mcp-demo/dynamic-headers/client\": \"Forward per-run user and session context from AgentOS to an MCP server with dynamic HTTP headers.\",\n  \"examples/agent-os/mcp-demo/oauth-authkit-example\": \"Delegate MCP OAuth to WorkOS AuthKit by passing its external AuthProvider as mcp_auth.\",\n  \"examples/agent-os/mcp-demo/oauth-builtin-example\": \"Run AgentOS as the OAuth authorization server for /mcp, with access gated by a deployer connect secret.\",\n  \"examples/agent-os/mcp-demo/mcp-tools-advanced-example\": \"Run an AgentOS support agent wired to both a remote Agno docs MCP server and a stdio Brave Search MCP server.\",\n  \"examples/agent-os/mcp-demo/mcp-tools-existing-lifespan\": \"Pass a custom FastAPI lifespan to AgentOS alongside agent MCPTools, whose lifespan AgentOS still manages.\",\n  \"examples/agent-os/mcp-demo/mcp-server-example\": \"Enable an LLM-friendly MCP server at /mcp on an AgentOS serving a web research agent via mcp_server=True.\",\n  \"examples/agent-os/middleware/agent-os-with-custom-middleware\": \"Attach per-IP rate limiting and request/response logging middleware to an AgentOS FastAPI app.\",\n  \"examples/agent-os/middleware/agent-os-with-jwt-middleware\": \"Inject user_id, session_id, and dependency claims from an Authorization-header JWT into AgentOS agent runs.\",\n  \"examples/agent-os/middleware/custom-fastapi-app-with-jwt-middleware\": \"Protect a custom FastAPI app mounted as AgentOS base_app with JWT auth, excluding a /auth/login token-issuing route.\",\n  \"examples/agent-os/middleware/extract-content-middleware\": \"Capture the X-APP-UUID header and extract agent response content from streaming and non-streaming /runs responses with a Starlette middleware.\",\n  \"examples/agent-os/middleware/guardrails-demo\": \"Attach OpenAI moderation, prompt-injection, and PII detection guardrails as pre-hooks on an AgentOS agent and team.\",\n  \"examples/agent-os/middleware/agent-os-with-service-accounts\": \"Mint, authenticate with, and revoke opaque agno_pat_ service-account tokens on AgentOS using an admin-scoped JWT.\",\n  \"examples/agent-os/rbac/asymmetric/custom-scope-mappings\": \"Attach JWTMiddleware with RS256 keys and a per-endpoint scope_mappings table so routes such as POST /agents/*/runs require multiple custom scopes.\",\n  \"examples/agent-os/rbac/symmetric/advanced-scopes\": \"Issue HS256 tokens across five privilege tiers to show global, per-agent, and wildcard scopes filtering the agent list and gating agent runs.\",\n  \"examples/agent-os/rbac/symmetric/basic\": \"Protect AgentOS endpoints with HS256 JWTs via AuthorizationConfig and default scope mappings, minting user and admin test tokens.\",\n  \"examples/agent-os/rbac/symmetric/user-isolation\": \"Scope non-admin JWT callers to their own sessions, memories, and traces with AuthorizationConfig(user_isolation=True), while agent_os:admin bypasses isolation.\",\n  \"examples/agent-os/remote/agent-os-gateway\": \"Front a single AgentOS gateway over remote AgentOS, Agno A2A, and Google ADK agents plus a local conditional story workflow.\",\n  \"examples/agent-os/remote/remote-adk-agent\": \"Call a Google ADK facts agent over A2A JSON-RPC with RemoteAgent, covering single-shot, streaming, and agent-card lookup.\",\n  \"examples/agent-os/remote/remote-agno-a2a-agent\": \"Call Agno agents exposed over the A2A REST interface with RemoteAgent, covering single-shot, streaming, and agent-card lookup.\",\n  \"examples/agent-os/remote/remote-agent-as-team-member\": \"Mix a local Agent with two RemoteAgents from an AgentOS on port 7778 in one hybrid Team.\",\n  \"examples/agent-os/remote/a2a-agent-as-team-member\": \"Add an A2A-protocol RemoteAgent (REST transport) alongside a local Agent in one cross-framework Team.\",\n  \"examples/agent-os/scheduler/basic-schedule\": \"Runs an AgentOS on port 7777 with cron scheduling and Postgres, then creates a 5-minute greeter schedule via POST /schedules.\",\n  \"examples/agent-os/scheduler/schedule-management\": \"Walks the full /schedules REST lifecycle with httpx against a running AgentOS: create, list, patch, disable/enable, trigger, read run history, delete.\",\n  \"examples/agent-os/scheduler/scheduler-tools-agent\": \"Give an AgentOS agent SchedulerTools to create, inspect, enable or disable, and delete cron schedules and view run history.\",\n  \"examples/agent-os/schemas/agent-schemas\": \"Serve two AgentOS agents: a HackerNews agent with a Pydantic input_schema and a movie agent with a MovieScript output_schema.\",\n  \"examples/agent-os/schemas/team-schemas\": \"Serve two AgentOS teams: one validating requests with a Pydantic input_schema, one returning a ResearchReport output_schema.\",\n  \"examples/agent-os/tracing/advanced-trace-filtering\": \"Build AND/OR/NOT, CONTAINS, IN and range trace queries with the FilterExpr DSL and run them against SqliteDb.get_traces.\",\n  \"examples/agent-os/workflow/basic-chat-workflow-agent\": \"Serve a story-writing workflow on AgentOS where a WorkflowAgent decides when to run write, conditional-edit, format, and reference steps backed by Postgres.\",\n  \"examples/agent-os/workflow/workflow-with-custom-function\": \"Runs an AgentOS content workflow that toggles between a sync custom-function executor (Postgres) and a streaming async one (SQLite).\",\n  \"examples/agent-os/workflow/workflow-with-workflow-as-step\": \"Nest three levels of workflows, with a Parallel step wrapping a Condition-gated fact check, served through AgentOS on Postgres.\",\n  \"examples/agent-os/agno-assist\": \"Serve a Claude-powered Agno docs assistant that queries the Agno docs MCP server and keeps SQLite-backed history.\",\n  \"examples/agent-os/factories/agent/basic-factory\": \"Serve an AgentFactory that builds a fresh per-tenant Agent from RequestContext.user_id on every request, alongside a static prototype agent, backed by PostgresDb.\",\n  \"examples/agent-os/factories/agent/input-schema-factory\": \"Attach a pydantic input_schema to an AgentFactory so AgentOS validates the client's factory_input (persona, depth) and exposes it as ctx.input when building the research agent.\",\n  \"examples/agent-os/factories/agent/hitl-factory\": \"AgentFactory whose Hacker News tool is marked requires_confirmation, pausing the run until it is approved or rejected via the /continue endpoint.\",\n  \"examples/agent-os/factories/workflow/basic-workflow-factory\": \"WorkflowFactory that rebuilds a two-step draft-then-edit content pipeline per request with tenant-specific agent instructions.\",\n  \"examples/agent-os/file-generation/file-generation-os\": \"Serve an agent with FileGenerationTools through AgentOS to produce JSON, CSV, PDF, DOCX, TXT, and HTML files returned as base64 artifacts.\",\n  \"examples/agent-os/human-in-the-loop/agent/agent-tool-requires-confirmation\": \"Pause an AgentOS agent run for user confirmation before its Hacker News fetch tool executes, using @tool(requires_confirmation=True).\",\n  \"examples/agent-os/human-in-the-loop/team/confirmation-required\": \"Pause a team run when a member's weather tool requires confirmation, then resume it via continue_run through AgentOS.\",\n  \"examples/agent-os/human-in-the-loop/team/external-tool-execution\": \"Team member tool marked external_execution so AgentOS pauses the run and the client executes send_email and returns the result via continue_run.\",\n  \"examples/agent-os/human-in-the-loop/team/team-tool-confirmation\": \"Team-level approve_deployment tool with requires_confirmation, pausing the AgentOS run until the client confirms or rejects the deployment.\",\n  \"examples/agent-os/human-in-the-loop/team/user-input-required\": \"Team member's plan_trip tool declares requires_user_input for destination and budget, pausing the AgentOS run until the client supplies the missing fields.\",\n  \"examples/agent-os/human-in-the-loop/workflow/condition-user-decision\": \"Workflow Condition with requires_confirmation where approving runs the detailed-analysis branch and rejecting routes to the quick-summary else branch.\",\n  \"examples/agent-os/human-in-the-loop/workflow/decision-tree\": \"Chain confirmation Conditions for analysis depth and output format in one AgentOS workflow.\",\n  \"examples/agent-os/human-in-the-loop/workflow/dual-level-hitl\": \"One workflow step pauses twice: first for step-level user input of the destination city, then for confirmation of the agent's book_flight tool call.\",\n  \"examples/reasoning/agents/analyse-treaty-of-versailles\": \"Compare gpt-4o with reasoning=True against a DeepSeek reasoner reasoning_model on a nuanced Treaty of Versailles analysis, streaming full reasoning.\",\n  \"examples/reasoning/agents/cerebras-llama-default-cot\": \"Stream a Cerebras agent's default chain-of-thought with reasoning and debug mode enabled.\",\n  \"examples/reasoning/agents/default-chain-of-thought\": \"Contrasts an explicit gpt-4o reasoning_model fallback against reasoning=True built-in chain-of-thought on the same fibonacci prompt.\",\n  \"examples/reasoning/agents/fibonacci\": \"Compares gpt-4o built-in chain-of-thought with a deepseek-reasoner reasoning model on a fibonacci scripting prompt.\",\n  \"examples/reasoning/agents/finance-agent\": \"Compares built-in chain-of-thought with a deepseek-reasoner reasoning model on gpt-4o YFinance agents writing an NVDA vs TSLA report.\",\n  \"examples/reasoning/agents/ibm-watsonx-default-cot\": \"Runs an IBM WatsonX llama-3-3-70b-instruct agent with reasoning=True to stream default chain-of-thought steps in debug mode.\",\n  \"examples/reasoning/agents/life-in-500000-years\": \"Compare gpt-4o built-in chain-of-thought against a deepseek-reasoner reasoning model on a 500,000-year sci-fi short story, streaming full reasoning for both.\",\n  \"examples/reasoning/agents/logical-puzzle\": \"Solve the missionaries-and-cannibals river crossing with an ASCII diagram, contrasting gpt-4o built-in chain-of-thought with a deepseek-reasoner reasoning model.\",\n  \"examples/reasoning/agents/mathematical-proof\": \"Prove that the sum of the first n odd numbers equals n squared, comparing gpt-4o built-in chain-of-thought with a deepseek-reasoner reasoning model.\",\n  \"examples/reasoning/agents/mistral-reasoning-cot\": \"Run mistral-large-latest with reasoning=True and use_json_mode=True, streaming the full chain-of-thought for a Fibonacci script walkthrough.\",\n  \"examples/reasoning/agents/python-101-curriculum\": \"Draft a Python 101 curriculum twice, once with gpt-4o built-in chain-of-thought and once with a deepseek-reasoner reasoning model, streaming full reasoning.\",\n  \"examples/reasoning/agents/scientific-research\": \"Critique a small-sample teaching-method abstract for bias and flawed conclusions, comparing gpt-4o built-in chain-of-thought with a deepseek-reasoner reasoning model.\",\n  \"examples/reasoning/agents/ship-of-theseus\": \"Compare an agent with built-in chain-of-thought against one using deepseek-reasoner on the Ship of Theseus identity debate.\",\n  \"examples/reasoning/agents/strawberry\": \"Run three async agents (plain, built-in chain-of-thought, deepseek-reasoner) on the 'r' in 'strawberry' count, with rich console rules.\",\n  \"examples/reasoning/agents/trolley-problem\": \"Solve the trolley problem with ASCII diagrams using a built-in chain-of-thought agent and a deepseek-reasoner-backed agent.\",\n  \"examples/evals/accuracy/accuracy-9-11-bigger-or-9-99\": \"Score an agent that must use CalculatorTools to answer whether 9.11 or 9.9 is bigger, with an o4-mini judge.\",\n  \"examples/evals/accuracy/accuracy-basic\": \"Run AccuracyEval with run() and arun() over a multi-step calculator prompt, using 1 and 3 iterations.\",\n  \"examples/evals/accuracy/accuracy-team\": \"Score a two-agent English/Spanish routing Team on its refusal response to a French prompt.\",\n  \"examples/evals/accuracy/accuracy-with-given-answer\": \"Grade a pre-computed output string with AccuracyEval.run_with_output(), no agent execution required.\",\n  \"examples/evals/accuracy/accuracy-with-tools\": \"Score a CalculatorTools-equipped agent on computing 10! against the expected 3628800.\",\n  \"examples/evals/accuracy/db-logging\": \"Persist AccuracyEval runs to a PostgresDb eval_runs_cookbook table while scoring a calculator agent.\",\n  \"examples/evals/accuracy/evaluator-agent\": \"Score an agent's step-by-step calculator math against an expected answer using a custom evaluator agent with an AccuracyAgentResponse output schema.\",\n  \"examples/evals/accuracy/accuracy-eval-metrics\": \"Accumulate evaluator token usage into an agent's run_output via the run_metrics parameter of evaluate_answer, surfacing it under the eval_model key of metrics.details.\",\n  \"examples/evals/agent-as-judge/agent-as-judge-basic\": \"Run numeric-scored agent-as-judge evaluations synchronously against PostgresDb and asynchronously against AsyncSqliteDb, with an on_fail callback and stored eval runs.\",\n  \"examples/evals/agent-as-judge/agent-as-judge-batch\": \"Judge three pre-supplied customer-service cases with binary scoring, persist the evaluation runs in SqliteDb, and report the result pass rate.\",\n  \"examples/evals/agent-as-judge/agent-as-judge-binary\": \"Judge a customer-service agent reply as pass or fail against a professional-tone criterion using AgentAsJudgeEval backed by SqliteDb.\",\n  \"examples/evals/agent-as-judge/agent-as-judge-custom-evaluator\": \"Score agent output with AgentAsJudgeEval using a strict custom evaluator_agent instead of the default judge.\",\n  \"examples/evals/agent-as-judge/agent-as-judge-post-hook\": \"Attach AgentAsJudgeEval as an agent post_hook and read stored scores back from SqliteDb and AsyncSqliteDb.\",\n  \"examples/evals/agent-as-judge/agent-as-judge-team\": \"Judge a researcher/writer Team response with binary AgentAsJudgeEval scoring persisted to SqliteDb.\",\n  \"examples/evals/agent-as-judge/agent-as-judge-team-post-hook\": \"Register AgentAsJudgeEval in Team.post_hooks to score collaboration quality automatically after each team run.\",\n  \"examples/evals/agent-as-judge/agent-as-judge-with-guidelines\": \"Steer numeric AgentAsJudgeEval scoring with additional_guidelines that require units, variant context, and technical completeness.\",\n  \"examples/evals/agent-as-judge/agent-as-judge-with-tools\": \"Score a CalculatorTools agent's math answer on step-by-step clarity with numeric AgentAsJudgeEval.\",\n  \"examples/evals/agent-as-judge/agent-as-judge-eval-metrics\": \"Attach AgentAsJudgeEval as an agent post_hook and read the evaluator's token usage from run_output.metrics.details['eval_model'].\",\n  \"examples/evals/performance/async-function\": \"Benchmark an async agent.arun call over 10 iterations with PerformanceEval.arun, printing runtime and memory results.\",\n  \"examples/evals/performance/comparison/autogen-instantiation\": \"Measure AutoGen AssistantAgent instantiation over 1000 iterations with PerformanceEval, using a gpt-4o OpenAIChatCompletionClient and a weather tool.\",\n  \"examples/evals/performance/comparison/crewai-instantiation\": \"Measure CrewAI Agent instantiation over 1000 iterations with PerformanceEval, using a gpt-4o llm and a decorated weather tool.\",\n  \"examples/evals/performance/comparison/langgraph-instantiation\": \"Measure LangGraph create_react_agent instantiation over 1000 iterations with PerformanceEval, using ChatOpenAI gpt-4o and a weather tool.\",\n  \"examples/evals/performance/comparison/openai-agents-instantiation\": \"Benchmark 1000 OpenAI Agents SDK agent constructions with a function tool using Agno's PerformanceEval.\",\n  \"examples/evals/performance/comparison/pydantic-ai-instantiation\": \"Benchmark 1000 PydanticAI agent constructions, including an inline @agent.tool_plain weather tool, via Agno's PerformanceEval.\",\n  \"examples/evals/performance/comparison/smolagents-instantiation\": \"Measure runtime and memory across 1,000 Smolagents ToolCallingAgent constructions per metric with PerformanceEval.\",\n  \"examples/evals/performance/db-logging\": \"Run a single-iteration agent run benchmark and persist its PerformanceEval results to the eval_runs_cookbook table in PostgresDb.\",\n  \"examples/evals/performance/instantiate-agent\": \"Measure runtime and memory of 1000 bare Agno Agent constructions with only a system message.\",\n  \"examples/evals/performance/instantiate-agent-with-tool\": \"Measure runtime and memory of 1000 Agno Agent constructions using OpenAIChat gpt-4o plus a weather tool.\",\n  \"examples/evals/performance/instantiate-team\": \"Benchmark Team construction cost over 1000 iterations with PerformanceEval.\",\n  \"examples/evals/performance/response-with-memory-updates\": \"Benchmark agent run latency with update_memory_on_run enabled against a SqliteDb over 5 iterations.\",\n  \"examples/evals/performance/response-with-storage\": \"Benchmark two chained agent runs with SqliteDb-backed history added to context.\",\n  \"examples/evals/performance/team-response-with-memory-and-reasoning\": \"Track memory growth and top allocations for a PostgresDb-backed team using ReasoningTools across five concurrent users.\",\n  \"examples/evals/performance/team-response-with-memory-multi-user\": \"Track memory growth for a PostgresDb-backed team answering five users concurrently, each in its own session.\",\n  \"examples/evals/performance/team-response-with-memory-simple\": \"Benchmark memory growth across 5 async runs of a weather team with persistent memory and history in Postgres.\",\n  \"examples/evals/reliability/db-logging\": \"Persist ReliabilityEval tool-call results to a Postgres eval_runs table via PostgresDb.\",\n  \"examples/evals/reliability/multiple-tool-calls/calculator\": \"Assert an agent calls both multiply and exponentiate, then relax to subset matching with allow_additional_tool_calls.\",\n  \"examples/evals/reliability/reliability-async\": \"Score a factorial tool call using ReliabilityEval.arun() driven by asyncio.run.\",\n  \"examples/evals/reliability/single-tool-calls/calculator\": \"Assert a single expected tool call and validate its arguments with expected_tool_call_arguments.\",\n  \"examples/evals/suite/suite-team-scoring\": \"Run eval Cases against a Team and grade each answer with a numeric 1-10 judge and threshold.\",\n  \"examples/integrations/memory/mem0-integration\": \"Seed a Mem0 cloud memory client with user facts and inject the retrieved memories into an Agno agent's context via dependencies.\",\n  \"examples/integrations/memory/memori-integration\": \"Register Memori against the agent's OpenAI client with a SQLite-backed store so preferences from earlier turns are recalled in later ones.\",\n  \"examples/integrations/memory/zep-integration\": \"Write user messages through ZepTools, wait for Zep to sync, then pass the retrieved Zep context block into an Agno agent as a dependency.\",\n  \"examples/memory/integrations/dakera-integration\": \"Pinned Dakera memory integration that requires current API migration and sends recalled context to OpenAI.\",\n  \"examples/integrations/observability/agent-ops\": \"Initialize AgentOps and auto-log a GPT-4o Agno agent's model calls.\",\n  \"examples/integrations/observability/arize-phoenix-moving-traces-to-different-projects\": \"Route traces from a stock agent and a search agent into separate Phoenix projects using dangerously_using_project.\",\n  \"examples/integrations/observability/arize-phoenix-via-openinference\": \"Trace a structured-output stock agent to Phoenix Cloud via phoenix.otel register with auto-instrumentation.\",\n  \"examples/integrations/observability/arize-phoenix-via-openinference-local\": \"Send stock-agent traces to a self-hosted Phoenix collector at localhost:6006 under a named project.\",\n  \"examples/integrations/observability/atla-op\": \"Wrap a web-search agent run in atla_insights instrument_agno to capture OpenAI traces in Atla.\",\n  \"examples/integrations/observability/langfuse-via-openinference\": \"Export Agno agent spans to Langfuse over OTLP HTTP with base64 basic-auth headers and AgnoInstrumentor.\",\n  \"examples/integrations/observability/langfuse-via-openinference-response-model\": \"Trace a YFinance stock-price agent with a Pydantic output_schema to Langfuse via the OpenInference Agno instrumentor and OTLP HTTP export.\",\n  \"examples/integrations/observability/langfuse-via-openlit\": \"Export traces from a web-search agent to Langfuse by wiring an OTLP HTTP span exporter into OpenLIT's global tracer with batching disabled.\",\n  \"examples/integrations/observability/langsmith-via-openinference\": \"Send spans from a web-search stock-news agent to the LangSmith EU OTLP endpoint using the OpenInference Agno instrumentor with API-key and project headers.\",\n  \"examples/integrations/observability/langtrace-op\": \"Auto-instrument a YFinance stock-price agent with langtrace.init() from the Langtrace Python SDK.\",\n  \"examples/integrations/observability/langwatch-op\": \"Register the OpenInference AgnoInstrumentor through langwatch.setup() so a YFinance stock-price agent reports spans to LangWatch.\",\n  \"examples/integrations/observability/latitude-via-openinference\": \"Stream an async YFinance stock-price agent and export its OpenInference spans to Latitude's OTLP ingest endpoint via bearer-token and project headers.\",\n  \"examples/integrations/observability/logfire-via-openinference\": \"Streams a YFinance stock agent's spans to Logfire over OTLP HTTP using the OpenInference Agno instrumentor.\",\n  \"examples/integrations/observability/maxim-ops\": \"Traces an interactive web-search plus YFinance team chat loop with Maxim's instrument_agno logger.\",\n  \"examples/integrations/observability/opik-via-openinference\": \"Exports agent, model, and tool spans with custom trace attributes to Opik via an OTLP HTTP span processor.\",\n  \"examples/integrations/observability/teams/langfuse-via-openinference-team\": \"Traces a two-agent market-data and news team to Langfuse over OTLP with a run_mode switch between sync and async runs.\",\n  \"examples/integrations/observability/trace-to-database\": \"Persists agent spans to SQLite with setup_tracing, then queries db.get_trace and db.get_spans to print the span tree and attributes.\",\n  \"examples/integrations/observability/traceloop-op\": \"Wraps an agent run in a Traceloop @workflow-decorated function so the run appears under a parent workflow span.\",\n  \"examples/integrations/observability/weave-op\": \"Wrap an Agno agent run in a @weave.op() decorator to log calls to a Weave project.\",\n  \"examples/integrations/observability/workflows/arize-phoenix-via-openinference-workflow\": \"Trace a research-summarize-fact-check-write workflow to Arize Phoenix using phoenix.otel auto-instrumentation.\",\n  \"examples/integrations/observability/workflows/langfuse-via-openinference-workflows\": \"Export OpenTelemetry spans from a conditional research workflow to Langfuse via the OpenInference Agno instrumentor.\",\n  \"examples/integrations/observability/mlflow-via-autolog\": \"Enable mlflow.agno.autolog() against a local MLflow tracking server to capture traces from a YFinance stock agent.\",\n  \"examples/integrations/observability/mlflow-via-openinference\": \"Send OpenInference spans from an async YFinance agent to an MLflow tracking server over the OTLP HTTP trace endpoint.\",\n  \"examples/integrations/rag/agentic-rag-infinity-reranker\": \"Agentic RAG over Agno docs using LanceDB hybrid search, Cohere embeddings, and a local Infinity reranker server on port 7997.\",\n  \"examples/integrations/rag/agentic-rag-with-lightrag\": \"Build agentic RAG over a LightRag vector database with a PDF, a Wikipedia topic, and a URL in one knowledge base.\",\n  \"examples/integrations/parallel/research-assistant\": \"Persistent Parallel research agent combining Search, Extract, and Task APIs with SQLite-backed sessions, history, and user memory across follow-up turns.\",\n  \"examples/integrations/parallel/web-plus-knowledge\": \"Agent that routes between a local ChromaDB knowledge base (hybrid search, OpenAI embeddings) and Parallel live web search depending on whether the question needs internal or current information.\",\n  \"examples/integrations/parallel/research-team\": \"Agno Team pairing a Parallel Search/Extract web researcher with a Task-API deep researcher, whose lead synthesizes one cited answer.\",\n  \"examples/integrations/parallel/research-workflow\": \"Run source gathering and cited-brief writing as two defined workflow steps persisted to SQLite.\",\n  \"examples/integrations/parallel/competitive-intel-monitor\": \"Agent using the Parallel Monitor API at a 1-day frequency to create topic monitors, list active ones, and report detected change events with sources.\",\n  \"examples/components/agent-os-registry\": \"Serve an AgentOS app whose Registry pre-registers DuckDuckGo and calculator tools, a custom function, and OpenAI/Claude/Azure models.\",\n  \"examples/components/demo\": \"Register tools, custom functions, Pydantic schemas, models, and a PgVector vector DB in an AgentOS Registry.\",\n  \"examples/components/get-agent\": \"Fetch a stored agent from PostgresDb with get_agent_by_id and run it, with get_agents shown for listing all.\",\n  \"examples/components/get-team\": \"Fetch a stored team from PostgresDb with get_team_by_id and stream its response, with get_teams shown for listing all.\",\n  \"examples/components/get-workflow\": \"Fetch a stored workflow from PostgresDb with get_workflow_by_id and run it, with get_workflows shown for listing all.\",\n  \"examples/components/registry\": \"Register tools, models, dbs, and Pydantic schemas in a Registry so saved agents can be rehydrated from Postgres.\",\n  \"examples/components/save-agent\": \"Persist an agent to PostgresDb with agent.save(), which returns a new version number on each save.\",\n  \"examples/components/save-team\": \"Persist a two-member Team (researcher and writer) to PostgresDb with team.save(), returning a new version each time.\",\n  \"examples/components/save-workflow\": \"Persist a two-step content workflow to PostgresDb with workflow.save(db=db), which versions the workflow on each save.\",\n  \"examples/components/workflows/save-conditional-steps\": \"Save a workflow whose Condition evaluator is a plain function, then reload it from Postgres by registering that function in a Registry.\",\n  \"examples/components/workflows/save-custom-steps\": \"Register a custom executor function in a Registry so a saved workflow's non-agent step resolves on load from Postgres.\",\n  \"examples/components/workflows/save-loop-steps\": \"Persist a Loop step whose end_condition function is restored from a Registry, iterating HackerNews and web research up to 3 times before summarizing.\",\n  \"examples/components/workflows/save-parallel-steps\": \"Save a Parallel research pipeline (HackerNews plus web) to Postgres and reload it by ID without a Registry, since all steps are agent-backed.\",\n  \"examples/components/workflows/save-router-steps\": \"Persist a Router whose keyword-based selector function is restored from a Registry, dispatching topics to HackerNews or web research before summarizing.\",\n  \"examples/components/workflows/registry-agents-in-workflow\": \"Expose code-defined agents with stable IDs through the AgentOS /registry endpoint so UI-built workflows resolve them by ID instead of loading them from the database.\",\n  \"examples/components/workflows/save-hitl-condition-loop-router\": \"Round-trip requires_confirmation, on_reject, and route-selection settings on Condition, Loop, and Router through workflow save/load, then drive the pauses interactively.\",\n  \"examples/components/workflows/save-hitl-confirmation-steps\": \"Save a workflow whose ProcessData step requires confirmation, reload it via get_workflow_by_id with a Registry, and confirm or skip the paused step at run time.\",\n  \"examples/components/workflows/save-hitl-user-input-steps\": \"Persist a workflow whose step declares a UserInputField schema (tone, length, include_examples), reload it from Postgres, and fill the paused step's fields interactively.\",\n  \"examples/components/auto-populate-registry-os\": \"Serve an AgentOS with no explicit registry and read the auto-discovered models, tools, and dbs from the GET /registry endpoint.\",\n  \"examples/models/openrouter/responses/stream\": \"Stream an OpenRouter Responses agent's output with print_response(stream=True).\",\n  \"examples/models/openrouter/responses/tool-use\": \"Call DuckDuckGoTools from an OpenRouter Responses agent and stream the answer.\",\n  \"examples/models/openrouter/chat/structured-output\": \"Compare JSON mode and native structured outputs on OpenRouter by generating a MovieScript Pydantic model.\",\n  \"examples/models/openrouter/chat/tool-use\": \"Give an OpenRouter agent WebSearchTools and stream the answer in both sync and async runs.\",\n  \"examples/tools/exceptions/retry-tool-call-from-post-hook\": \"Automatically retry a tool call from a post-hook with RetryAgentRun.\",\n  \"examples/tools/mcp/gibsonai\": \"Connect a local GibsonAI MCP server to an Agno agent.\",\n  \"examples/tools/mcp/supabase\": \"Use the Supabase MCP server to create projects, database schemas, and edge functions.\",\n  \"examples/tools/spider-tools\": \"Search, scrape, and crawl websites into LLM-ready Markdown with SpiderTools.\",\n  \"examples/tools/models/gemini-video-generation\": \"Migrate GeminiTools to Veo 3.1 on Vertex AI and save the returned MP4 correctly.\",\n  \"examples/tools/other/human-in-the-loop\": \"Require user confirmation in a tool pre-hook before fetching Hacker News stories.\",\n  \"examples/agent-os/client-a2a/servers/google-adk-server\": \"Expose a Google ADK search agent over A2A for A2AClient tests.\",\n  \"examples/agent-os/factories/hitl-factory\": \"Build a factory agent whose tool pauses for confirmation before execution.\",\n  \"examples/models/openrouter/responses/basic\": \"Run openai/gpt-oss-20b through OpenRouterResponses with reasoning enabled in sync and async modes.\",\n  \"examples/models/openrouter/responses/fallback\": \"Configure ordered OpenRouter fallback models for unavailable primary routes.\",\n  \"examples/tools/mcp/cli\": \"Run an interactive CLI chat loop against a GitHub MCP server agent.\",\n  \"examples/tools/mcp/local-server/server\": \"Serve weather tools from a minimal FastMCP stdio server.\",\n  \"examples/tools/mcp/notion-mcp-agent\": \"Query and update connected Notion pages through the official Notion MCP server.\",\n  \"examples/tools/browserbase-tools\": \"Navigate quotes.toscrape.com and extract paginated quotes and authors with BrowserbaseTools.\",\n  \"examples/tools/e2b-tools\": \"Run Python, manage files, and expose sandbox servers with E2BTools.\",\n  \"examples/workflows/human-in-the-loop/dual-level-hitl/multi-step-mixed-hitl\": \"Combine step confirmation, user input, and executor tool confirmation in one streaming workflow.\",\n  \"examples/workflows/human-in-the-loop/dual-level-hitl/output-review-and-tool-confirmation\": \"Confirm the agent's tool call during execution, then review the step output after it completes.\",\n  \"examples/workflows/human-in-the-loop/dual-level-hitl/loop-confirmation-and-tool-confirmation\": \"Confirm the Loop before it starts, then confirm the agent's tool call on each iteration.\",\n  \"examples/workflows/human-in-the-loop/dual-level-hitl/router-confirmation-and-tool-confirmation\": \"Confirm the Router before its selector runs, then confirm the tool call made by the agent on the chosen route.\",\n  \"examples/context/wiki-notion\": \"Mirror a Notion database to local Markdown and update it through WikiContextProvider.\",\n  \"examples/agent-os/dbs/neon\": \"Use a NEON_DB_URL-backed PostgresDb for AccuracyEval in an AgentOS with an agent and team.\",\n  \"examples/tools/clickup-tools\": \"List ClickUp spaces and manage tasks with ClickUpTools using CLICKUP_API_KEY and MASTER_SPACE_ID.\",\n  \"examples/tools/dalle-tools\": \"Legacy DalleTools reference for configuring deprecated DALL-E models.\",\n  \"examples/tools/mcp/sequential-thinking\": \"Combine the sequential-thinking MCP server with YFinanceTools to reason step by step over stock data.\",\n  \"examples/agent-os/interfaces/slack/multimodal-team\": \"Legacy Slack team example with image analysis, web research, and DalleTools generation.\",\n  \"examples/agent-os/interfaces/slack/multimodal-workflow\": \"Legacy Slack workflow example with parallel media analysis and DalleTools generation.\",\n  \"examples/agent-os/interfaces/whatsapp/multimodal-team\": \"Legacy WhatsApp team example with media analysis, web research, and DalleTools generation.\",\n  \"examples/agent-os/interfaces/whatsapp/multimodal-workflow\": \"Legacy WhatsApp workflow example with parallel media analysis and DalleTools generation.\",\n  \"examples/agent-os/interfaces/telegram/agent-with-media\": \"Legacy Telegram media bot with DalleTools generation, ElevenLabs audio, and multimodal analysis.\",\n  \"examples/teams/multimodal/generate-image-with-team\": \"Legacy DalleTools team example for prompt refinement and image generation.\",\n  \"examples/models/google/gemini/parallel-grounding\": \"Ground Gemini 3.5 Flash responses with Parallel web search on Vertex AI.\",\n  \"examples/models/deepseek/thinking-mode\": \"Enable or disable DeepSeek V4 thinking mode and inspect reasoning_content.\",\n  \"examples/tools/google/workspace/service-account\": \"Authenticate Gmail, Calendar, Drive, and Sheets toolkits with a delegated Google Workspace service account.\"\n}\n"}}}}