1 line
22 KiB
JSON
1 line
22 KiB
JSON
{"repo": "bagros/mcp-research-router", "files": {"USAGE.md": {"cb4669cca2a55b70497033b1534998013fceec1a": {"t": 1785747202.8095345, "n": 10386, "h": "68558a57b2833a43", "c": "# MCP Research Router - Usage Guide\n\n## Quick Start\n\nThe simplest way to use MCP Research Router is via `npx` with Claude Desktop or any MCP-compatible client.\n\n## Installation Methods\n\n### Method 1: Via npx (Recommended)\n\nNo installation needed - run directly:\n\n```bash\nnpx @bagros/mcp-research-router@latest\n```\n\n### Method 2: Global Installation\n\n```bash\nnpm install -g @bagros/mcp-research-router\nmcp-research-router\n```\n\n### Method 3: Local Project\n\n```bash\nnpm install @bagros/mcp-research-router\nnpx mcp-research-router\n```\n\n## Claude Desktop Integration\n\n### Configuration File Location\n\n**Windows**:\n```\n%APPDATA%\\Claude\\claude_desktop_config.json\n```\n\n**macOS**:\n```\n~/Library/Application Support/Claude/claude_desktop_config.json\n```\n\n**Linux**:\n```\n~/.config/Claude/claude_desktop_config.json\n```\n\n### Basic Configuration\n\nEdit `claude_desktop_config.json`:\n\n```json\n{\n \"mcpServers\": {\n \"research-router\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"@bagros/mcp-research-router@latest\"],\n \"env\": {\n \"OPENAI_API_KEY\": \"sk-proj-...\",\n \"GOOGLE_API_KEY\": \"AIza...\",\n \"PERPLEXITY_API_KEY\": \"pplx-...\",\n \"DEEPSEEK_API_KEY\": \"sk-...\"\n }\n }\n }\n}\n```\n\n### Configuration with Specific Providers\n\nEnable only selected providers:\n\n```json\n{\n \"mcpServers\": {\n \"research-router\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"@bagros/mcp-research-router@latest\"],\n \"env\": {\n \"OPENAI_API_KEY\": \"sk-proj-...\",\n \"PERPLEXITY_API_KEY\": \"pplx-...\"\n }\n }\n }\n}\n```\n\nThe router will automatically skip providers without API keys.\n\n### Restart Claude Desktop\n\nAfter editing configuration:\n\n1. **Quit Claude Desktop completely**\n2. Restart the application\n3. Check for the research-router tools in the tool menu\n\n## Available Tools\n\n### 1. research.run\n\nExecute parallel research queries across multiple AI providers.\n\n**Parameters**:\n\n- `providers` (required): Array of provider names\n - Options: `\"openai\"`, `\"gemini\"`, `\"perplexity\"`, `\"deepseek\"`\n - Example: `[\"openai\", \"perplexity\"]`\n\n- `questions` (required): Array of research questions\n - Example: `[\"What is quantum computing?\", \"Latest AI breakthroughs?\"]`\n\n- `models` (optional): Provider-specific models\n - Example: `{ \"openai\": \"gpt-5-mini\", \"gemini\": \"gemini-2.5-flash\" }`\n\n- `synthesis` (optional): LLM model for synthesis\n - Default: `\"gpt-5-mini\"`\n - Options: `\"gpt-5\"`, `\"gpt-5-mini\"`, `\"gpt-5-nano\"`\n\n- `return` (optional): Output format\n - Options: `\"json\"` (default), `\"markdown\"`, `\"file\"`\n\n- `save` (optional): Auto-save to file\n - Type: `boolean`\n - Default: `false`\n\n**Example Usage in Claude**:\n\n```\nUse research.run to search for \"latest developments in quantum computing\" \nusing OpenAI and Perplexity, then synthesize the results.\n```\n\nClaude will call the tool with appropriate parameters.\n\n### 2. research.status\n\nGet current runtime metrics and statistics.\n\n**Returns**:\n- Total queries executed\n- Success/failure counts per provider\n- Average latency per provider\n- Total estimated costs\n- Rate limiter queue status\n\n**Example**:\n\n```\nCheck the research router status\n```\n\n### 3. research.models\n\nList available models for each provider.\n\n**Parameters**:\n\n- `provider` (optional): Specific provider name\n - If omitted, lists models for all providers\n\n**Example**:\n\n```\nShow me available models for Gemini\n```\n\n### 4. research.save\n\nManually save the last research results to a Markdown file.\n\n**Parameters**:\n\n- `filepath` (optional): Custom file path\n - Default: `./reports/research-{timestamp}.md`\n\n**Example**:\n\n```\nSave the last research results to a file\n```\n\n## Usage Examples\n\n### Example 1: Basic Research Query\n\n**Prompt to Claude**:\n\n```\nResearch \"What are the latest trends in AI safety?\" using OpenAI and Perplexity\n```\n\n**What happens**:\n1. Claude calls [`research.run`](src/tools/researchRun.ts:1)\n2. Query sent to both providers in parallel\n3. Results synthesized using GPT-5-mini\n4. Summary returned to Claude\n5. Claude presents findings to you\n\n### Example 2: Multi-Question Research\n\n**Prompt**:\n\n```\nI need research on three topics:\n1. Current state of quantum computing\n2. Recent breakthroughs in fusion energy\n3. Progress in carbon capture technology\n\nUse all available providers and save the results to a file\n```\n\n**Result**:\n- 12 parallel queries (4 providers × 3 questions)\n- Synthesized comprehensive report\n- Saved to `./reports/research-2025-10-26.md`\n\n### Example 3: Provider Comparison\n\n**Prompt**:\n\n```\nCompare how different AI providers answer: \"What is the capital of France?\"\nUse OpenAI, Gemini, Perplexity, and DeepSeek\n```\n\n**Result**:\n- Shows answer from each provider\n- Highlights any differences or contradictions\n- Useful for validating information\n\n### Example 4: Cost-Effective Research\n\n**Prompt**:\n\n```\nResearch \"machine learning basics\" using only the most cost-effective model\n```\n\nClaude will configure:\n```json\n{\n \"providers\": [\"gemini\"],\n \"models\": { \"gemini\": \"gemini-2.5-flash\" },\n \"synthesis\": \"gpt-5-nano\"\n}\n```\n\n## Output Formats\n\n### JSON Output (Default)\n\n```json\n{\n \"summary\": \"Synthesized findings...\",\n \"sources\": [\n {\n \"provider\": \"perplexity\",\n \"model\": \"sonar-pro\",\n \"answer\": \"...\",\n \"citations\": [\"https://...\"]\n }\n ],\n \"metadata\": {\n \"totalLatencyMs\": 5432,\n \"costEstimateUSD\": {\n \"openai\": 0.002,\n \"perplexity\": 0.004\n }\n }\n}\n```\n\n### Markdown Output\n\n```markdown\n# Research Summary: Quantum Computing\n\n*Generated: 2025-10-26T10:30:00Z*\n\n## Key Findings\n\n1. **Current State**: Quantum computers with 1000+ qubits...\n2. **Recent Breakthroughs**: Google's quantum error correction...\n\n## Sources\n\n- **Perplexity (sonar-pro)**: [Link](https://...)\n- **OpenAI (gpt-5)**: Analysis of...\n\n## Contradictions\n\n- Perplexity reported 1000 qubits\n- DeepSeek mentioned 1121 qubits\n```\n\n### File Output\n\nSaved to `./reports/research-2025-10-26T10-30-00.md` with:\n- YAML front matter (metadata)\n- Human-readable summary\n- Provenance for all sources\n- Cost and latency metrics\n\n## Advanced Usage\n\n### Rate Limiting\n\nThe router automatically manages:\n- **Global limit**: 6 concurrent requests\n- **Per-provider limit**: 2 requests/second\n\nNo configuration needed - handled automatically.\n\n### Cost Optimization\n\nEstimated costs per 1000 tokens:\n\n| Provider | Model | Cost |\n|----------|-------|------|\n| OpenAI | gpt-5-nano | $0.001 |\n| OpenAI | gpt-5-mini | $0.003 |\n| OpenAI | gpt-5 | $0.015 |\n| Gemini | gemini-2.5-flash | $0.000 (free tier) |\n| Perplexity | sonar | $0.001 |\n| Perplexity | sonar-pro | $0.003 |\n| DeepSeek | deepseek-chat | $0.001 |\n\n### Error Handling\n\nThe router handles:\n- **API failures**: Continues with other providers\n- **Rate limits**: Automatic backoff and retry\n- **Timeout**: 30-second default per request\n- **Invalid responses**: Logged and skipped\n\n### Logging\n\nAll events logged to:\n- **stderr**: Real-time console output\n- **File**: `./logs/research-router-YYYY-MM-DD.jsonl`\n\nLogs include:\n- Request/response timing\n- Cost estimates\n- Error details\n- Provider metrics\n\n## MCP Inspector Testing\n\nTest the server before using with Claude:\n\n```bash\n# Install MCP Inspector\nnpm install -g @modelcontextprotocol/inspector\n\n# Run with the research router\nnpx @modelcontextprotocol/inspector npx @bagros/mcp-research-router@latest\n```\n\nFeatures:\n- Interactive tool testing\n- Request/response inspection\n- Schema validation\n- Real-time logging\n\n## Troubleshooting\n\n### Tools Not Appearing in Claude\n\n1. Check configuration file location\n2. Verify JSON syntax (use validator)\n3. Ensure API keys are set\n4. Restart Claude Desktop completely\n5. Check Claude's MCP logs\n\n### API Key Errors\n\n```\nError: OPENAI_API_KEY not found\n```\n\n**Solution**: Add API key to `env` section in configuration.\n\n### Rate Limit Errors\n\n```\nError: Rate limit exceeded for provider: perplexity\n```\n\n**Solution**: Router handles automatically with backoff. If persistent, reduce concurrent requests.\n\n### Module Not Found\n\n```\nError: Cannot find module '@bagros/mcp-research-router'\n```\n\n**Solution**: \n- Check internet connection\n- Run: `npm cache clean --force`\n- Try: `npx -y @bagros/mcp-research-router@latest`\n\n### Synthesis Failures\n\n```\nError: Synthesis failed with gpt-5-mini\n```\n\n**Solution**: \n- Check OPENAI_API_KEY is valid\n- Verify OpenAI account has credits\n- Try different synthesis model\n\n## Best Practices\n\n### 1. Provider Selection\n\n- **Speed**: Gemini Flash (fastest)\n- **Citations**: Perplexity (best citations)\n- **Depth**: OpenAI GPT-5 (most detailed)\n- **Cost**: Gemini Flash (free tier)\n\n### 2. Question Formatting\n\n**Good**:\n```\n\"What are the latest developments in quantum computing as of 2025?\"\n```\n\n**Better**:\n```\n\"List the top 3 breakthrough achievements in quantum computing from 2024-2025, \nincluding organizations involved and qubit counts achieved\"\n```\n\n### 3. Synthesis Configuration\n\n- Use `gpt-5-nano` for simple summaries\n- Use `gpt-5-mini` for balanced analysis (default)\n- Use `gpt-5` for complex synthesis with deep analysis\n\n### 4. Cost Management\n\n```\nTotal = (Provider Queries × Cost) + (Synthesis Cost)\n```\n\nExample for 2 questions × 3 providers:\n- 6 queries @ $0.001 each = $0.006\n- 1 synthesis @ $0.003 = $0.003\n- **Total: ~$0.009**\n\n## API Key Management\n\n### Getting API Keys\n\n1. **OpenAI**: https://platform.openai.com/api-keys\n2. **Google AI Studio**: https://makersuite.google.com/app/apikey\n3. **Perplexity**: https://www.perplexity.ai/settings/api\n4. **DeepSeek**: https://platform.deepseek.com/api_keys\n\n### Security Best Practices\n\n- ✅ Store keys in configuration file (not in code)\n- ✅ Use environment-specific configurations\n- ✅ Rotate keys periodically\n- ✅ Monitor usage and costs\n- ❌ Never commit keys to version control\n- ❌ Don't share configuration files\n\n## Resources\n\n- **GitHub Repository**: https://github.com/BaGRoS/mcp-research-router\n- **npm Package**: https://www.npmjs.com/package/@bagros/mcp-research-router\n- **MCP Documentation**: https://modelcontextprotocol.io/\n- **Issue Tracker**: https://github.com/BaGRoS/mcp-research-router/issues\n\n## Support\n\nFor questions, issues, or feature requests:\n\n1. Check existing documentation\n2. Search GitHub issues\n3. Create new issue with:\n - Error message\n - Configuration (redact API keys)\n - Steps to reproduce\n\n---\n\n**Version**: 0.1.0 \n**Author**: Mirosław Bagrowski (@BaGRoS) \n**License**: MIT"}, "815261b4be9a24b3f66c4faddadfc55438eac562": {"t": 1785771350.0568135, "n": 10386, "h": "68558a57b2833a43", "c": "# MCP Research Router - Usage Guide\n\n## Quick Start\n\nThe simplest way to use MCP Research Router is via `npx` with Claude Desktop or any MCP-compatible client.\n\n## Installation Methods\n\n### Method 1: Via npx (Recommended)\n\nNo installation needed - run directly:\n\n```bash\nnpx @bagros/mcp-research-router@latest\n```\n\n### Method 2: Global Installation\n\n```bash\nnpm install -g @bagros/mcp-research-router\nmcp-research-router\n```\n\n### Method 3: Local Project\n\n```bash\nnpm install @bagros/mcp-research-router\nnpx mcp-research-router\n```\n\n## Claude Desktop Integration\n\n### Configuration File Location\n\n**Windows**:\n```\n%APPDATA%\\Claude\\claude_desktop_config.json\n```\n\n**macOS**:\n```\n~/Library/Application Support/Claude/claude_desktop_config.json\n```\n\n**Linux**:\n```\n~/.config/Claude/claude_desktop_config.json\n```\n\n### Basic Configuration\n\nEdit `claude_desktop_config.json`:\n\n```json\n{\n \"mcpServers\": {\n \"research-router\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"@bagros/mcp-research-router@latest\"],\n \"env\": {\n \"OPENAI_API_KEY\": \"sk-proj-...\",\n \"GOOGLE_API_KEY\": \"AIza...\",\n \"PERPLEXITY_API_KEY\": \"pplx-...\",\n \"DEEPSEEK_API_KEY\": \"sk-...\"\n }\n }\n }\n}\n```\n\n### Configuration with Specific Providers\n\nEnable only selected providers:\n\n```json\n{\n \"mcpServers\": {\n \"research-router\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"@bagros/mcp-research-router@latest\"],\n \"env\": {\n \"OPENAI_API_KEY\": \"sk-proj-...\",\n \"PERPLEXITY_API_KEY\": \"pplx-...\"\n }\n }\n }\n}\n```\n\nThe router will automatically skip providers without API keys.\n\n### Restart Claude Desktop\n\nAfter editing configuration:\n\n1. **Quit Claude Desktop completely**\n2. Restart the application\n3. Check for the research-router tools in the tool menu\n\n## Available Tools\n\n### 1. research.run\n\nExecute parallel research queries across multiple AI providers.\n\n**Parameters**:\n\n- `providers` (required): Array of provider names\n - Options: `\"openai\"`, `\"gemini\"`, `\"perplexity\"`, `\"deepseek\"`\n - Example: `[\"openai\", \"perplexity\"]`\n\n- `questions` (required): Array of research questions\n - Example: `[\"What is quantum computing?\", \"Latest AI breakthroughs?\"]`\n\n- `models` (optional): Provider-specific models\n - Example: `{ \"openai\": \"gpt-5-mini\", \"gemini\": \"gemini-2.5-flash\" }`\n\n- `synthesis` (optional): LLM model for synthesis\n - Default: `\"gpt-5-mini\"`\n - Options: `\"gpt-5\"`, `\"gpt-5-mini\"`, `\"gpt-5-nano\"`\n\n- `return` (optional): Output format\n - Options: `\"json\"` (default), `\"markdown\"`, `\"file\"`\n\n- `save` (optional): Auto-save to file\n - Type: `boolean`\n - Default: `false`\n\n**Example Usage in Claude**:\n\n```\nUse research.run to search for \"latest developments in quantum computing\" \nusing OpenAI and Perplexity, then synthesize the results.\n```\n\nClaude will call the tool with appropriate parameters.\n\n### 2. research.status\n\nGet current runtime metrics and statistics.\n\n**Returns**:\n- Total queries executed\n- Success/failure counts per provider\n- Average latency per provider\n- Total estimated costs\n- Rate limiter queue status\n\n**Example**:\n\n```\nCheck the research router status\n```\n\n### 3. research.models\n\nList available models for each provider.\n\n**Parameters**:\n\n- `provider` (optional): Specific provider name\n - If omitted, lists models for all providers\n\n**Example**:\n\n```\nShow me available models for Gemini\n```\n\n### 4. research.save\n\nManually save the last research results to a Markdown file.\n\n**Parameters**:\n\n- `filepath` (optional): Custom file path\n - Default: `./reports/research-{timestamp}.md`\n\n**Example**:\n\n```\nSave the last research results to a file\n```\n\n## Usage Examples\n\n### Example 1: Basic Research Query\n\n**Prompt to Claude**:\n\n```\nResearch \"What are the latest trends in AI safety?\" using OpenAI and Perplexity\n```\n\n**What happens**:\n1. Claude calls [`research.run`](src/tools/researchRun.ts:1)\n2. Query sent to both providers in parallel\n3. Results synthesized using GPT-5-mini\n4. Summary returned to Claude\n5. Claude presents findings to you\n\n### Example 2: Multi-Question Research\n\n**Prompt**:\n\n```\nI need research on three topics:\n1. Current state of quantum computing\n2. Recent breakthroughs in fusion energy\n3. Progress in carbon capture technology\n\nUse all available providers and save the results to a file\n```\n\n**Result**:\n- 12 parallel queries (4 providers × 3 questions)\n- Synthesized comprehensive report\n- Saved to `./reports/research-2025-10-26.md`\n\n### Example 3: Provider Comparison\n\n**Prompt**:\n\n```\nCompare how different AI providers answer: \"What is the capital of France?\"\nUse OpenAI, Gemini, Perplexity, and DeepSeek\n```\n\n**Result**:\n- Shows answer from each provider\n- Highlights any differences or contradictions\n- Useful for validating information\n\n### Example 4: Cost-Effective Research\n\n**Prompt**:\n\n```\nResearch \"machine learning basics\" using only the most cost-effective model\n```\n\nClaude will configure:\n```json\n{\n \"providers\": [\"gemini\"],\n \"models\": { \"gemini\": \"gemini-2.5-flash\" },\n \"synthesis\": \"gpt-5-nano\"\n}\n```\n\n## Output Formats\n\n### JSON Output (Default)\n\n```json\n{\n \"summary\": \"Synthesized findings...\",\n \"sources\": [\n {\n \"provider\": \"perplexity\",\n \"model\": \"sonar-pro\",\n \"answer\": \"...\",\n \"citations\": [\"https://...\"]\n }\n ],\n \"metadata\": {\n \"totalLatencyMs\": 5432,\n \"costEstimateUSD\": {\n \"openai\": 0.002,\n \"perplexity\": 0.004\n }\n }\n}\n```\n\n### Markdown Output\n\n```markdown\n# Research Summary: Quantum Computing\n\n*Generated: 2025-10-26T10:30:00Z*\n\n## Key Findings\n\n1. **Current State**: Quantum computers with 1000+ qubits...\n2. **Recent Breakthroughs**: Google's quantum error correction...\n\n## Sources\n\n- **Perplexity (sonar-pro)**: [Link](https://...)\n- **OpenAI (gpt-5)**: Analysis of...\n\n## Contradictions\n\n- Perplexity reported 1000 qubits\n- DeepSeek mentioned 1121 qubits\n```\n\n### File Output\n\nSaved to `./reports/research-2025-10-26T10-30-00.md` with:\n- YAML front matter (metadata)\n- Human-readable summary\n- Provenance for all sources\n- Cost and latency metrics\n\n## Advanced Usage\n\n### Rate Limiting\n\nThe router automatically manages:\n- **Global limit**: 6 concurrent requests\n- **Per-provider limit**: 2 requests/second\n\nNo configuration needed - handled automatically.\n\n### Cost Optimization\n\nEstimated costs per 1000 tokens:\n\n| Provider | Model | Cost |\n|----------|-------|------|\n| OpenAI | gpt-5-nano | $0.001 |\n| OpenAI | gpt-5-mini | $0.003 |\n| OpenAI | gpt-5 | $0.015 |\n| Gemini | gemini-2.5-flash | $0.000 (free tier) |\n| Perplexity | sonar | $0.001 |\n| Perplexity | sonar-pro | $0.003 |\n| DeepSeek | deepseek-chat | $0.001 |\n\n### Error Handling\n\nThe router handles:\n- **API failures**: Continues with other providers\n- **Rate limits**: Automatic backoff and retry\n- **Timeout**: 30-second default per request\n- **Invalid responses**: Logged and skipped\n\n### Logging\n\nAll events logged to:\n- **stderr**: Real-time console output\n- **File**: `./logs/research-router-YYYY-MM-DD.jsonl`\n\nLogs include:\n- Request/response timing\n- Cost estimates\n- Error details\n- Provider metrics\n\n## MCP Inspector Testing\n\nTest the server before using with Claude:\n\n```bash\n# Install MCP Inspector\nnpm install -g @modelcontextprotocol/inspector\n\n# Run with the research router\nnpx @modelcontextprotocol/inspector npx @bagros/mcp-research-router@latest\n```\n\nFeatures:\n- Interactive tool testing\n- Request/response inspection\n- Schema validation\n- Real-time logging\n\n## Troubleshooting\n\n### Tools Not Appearing in Claude\n\n1. Check configuration file location\n2. Verify JSON syntax (use validator)\n3. Ensure API keys are set\n4. Restart Claude Desktop completely\n5. Check Claude's MCP logs\n\n### API Key Errors\n\n```\nError: OPENAI_API_KEY not found\n```\n\n**Solution**: Add API key to `env` section in configuration.\n\n### Rate Limit Errors\n\n```\nError: Rate limit exceeded for provider: perplexity\n```\n\n**Solution**: Router handles automatically with backoff. If persistent, reduce concurrent requests.\n\n### Module Not Found\n\n```\nError: Cannot find module '@bagros/mcp-research-router'\n```\n\n**Solution**: \n- Check internet connection\n- Run: `npm cache clean --force`\n- Try: `npx -y @bagros/mcp-research-router@latest`\n\n### Synthesis Failures\n\n```\nError: Synthesis failed with gpt-5-mini\n```\n\n**Solution**: \n- Check OPENAI_API_KEY is valid\n- Verify OpenAI account has credits\n- Try different synthesis model\n\n## Best Practices\n\n### 1. Provider Selection\n\n- **Speed**: Gemini Flash (fastest)\n- **Citations**: Perplexity (best citations)\n- **Depth**: OpenAI GPT-5 (most detailed)\n- **Cost**: Gemini Flash (free tier)\n\n### 2. Question Formatting\n\n**Good**:\n```\n\"What are the latest developments in quantum computing as of 2025?\"\n```\n\n**Better**:\n```\n\"List the top 3 breakthrough achievements in quantum computing from 2024-2025, \nincluding organizations involved and qubit counts achieved\"\n```\n\n### 3. Synthesis Configuration\n\n- Use `gpt-5-nano` for simple summaries\n- Use `gpt-5-mini` for balanced analysis (default)\n- Use `gpt-5` for complex synthesis with deep analysis\n\n### 4. Cost Management\n\n```\nTotal = (Provider Queries × Cost) + (Synthesis Cost)\n```\n\nExample for 2 questions × 3 providers:\n- 6 queries @ $0.001 each = $0.006\n- 1 synthesis @ $0.003 = $0.003\n- **Total: ~$0.009**\n\n## API Key Management\n\n### Getting API Keys\n\n1. **OpenAI**: https://platform.openai.com/api-keys\n2. **Google AI Studio**: https://makersuite.google.com/app/apikey\n3. **Perplexity**: https://www.perplexity.ai/settings/api\n4. **DeepSeek**: https://platform.deepseek.com/api_keys\n\n### Security Best Practices\n\n- ✅ Store keys in configuration file (not in code)\n- ✅ Use environment-specific configurations\n- ✅ Rotate keys periodically\n- ✅ Monitor usage and costs\n- ❌ Never commit keys to version control\n- ❌ Don't share configuration files\n\n## Resources\n\n- **GitHub Repository**: https://github.com/BaGRoS/mcp-research-router\n- **npm Package**: https://www.npmjs.com/package/@bagros/mcp-research-router\n- **MCP Documentation**: https://modelcontextprotocol.io/\n- **Issue Tracker**: https://github.com/BaGRoS/mcp-research-router/issues\n\n## Support\n\nFor questions, issues, or feature requests:\n\n1. Check existing documentation\n2. Search GitHub issues\n3. Create new issue with:\n - Error message\n - Configuration (redact API keys)\n - Steps to reproduce\n\n---\n\n**Version**: 0.1.0 \n**Author**: Mirosław Bagrowski (@BaGRoS) \n**License**: MIT"}}}} |