{"repo": "cheneynine/papertracker", "files": {"src/PaperTracker/config/defaults.yml": {"4ad42526b0f226a619c059189aaae82c261d5644": {"t": 1785746339.6871982, "n": 4758, "h": "c67ab1b397b13d02", "c": "# Internal defaults.\n#\n# DO NOT edit this file to configure your instance.\n# To override any value, add the corresponding key to your own config file\n# (passed via --config). It will be deep-merged on top of these defaults.\n#\n# Required fields with no defaults — must be specified in user config:\n#   queries          — define at least one search query\n#   llm.base_url     — required when llm.enabled is true\n#   llm.model        — required when llm.enabled is true\n\n# runtime domain\nlog:\n  level: INFO\n  to_file: true\n  dir: log\n\n# dashboard domain\ndashboard:\n  host: 127.0.0.1\n  port: 8765\n  auto_refresh_seconds: 30\n\n# storage domain\nstorage:\n  enabled: true\n  db_path: database/papers.db     # Relative path (from working directory) or absolute path (starts with /)\n  content_storage_enabled: true   # save full paper content (default: true)\n  keep_arxiv_version: false       # false: 2601.21922, true: 2601.21922v1\n\n# Global scope (optional), applied to every query.\n# Field keys must be uppercase: TITLE/ABSTRACT/AUTHOR/JOURNAL/CATEGORY\n# Operator keys must be uppercase: AND/OR/NOT\n#\n# scope:\n#   CATEGORY:\n#     OR: [cs.CV]\n\nsearch:\n  sources: [arxiv]           # enabled search sources\n  max_results: 5              # max paper count for each query\n  ccf_enabled: false          # enable local CCF A/B whitelist helpers for supported sources\n  ccf_cache_path: database/ccf_venues_ab.json\n  ccf_ranks: [A, B]\n  dblp_recent_years: 2\n  openreview_recent_years: 2\n  openreview_max_pages: 3\n\n  # multi round fetch\n  pull_every: 7               # Strict time window, unit: day\n  fill_enabled: true          # Allow fill with paper outside the strict time window\n  max_lookback_days: 30       # Max length for fill window, -1 as infinite\n  max_fetch_items: 125        # Max number of items to fetch, -1 means unlimited\n  fetch_batch_size: 25        # Fetch count for each round\n  arxiv_min_interval_seconds: 12.0                  # Minimum spacing between arXiv requests across pages/retries\n  openalex_relevance_threshold: 0.0                    # local threshold for OpenAlex relevance_score\n\n# output domain\noutput:\n  base_dir: output/                      # Output root directory\n  formats: [console, json, html]         # Output formats: console, json, markdown, html (can select multiple)\n\n  # Markdown export configuration\n  markdown:\n    template_dir: template/markdown/\n    document_template: document.md       # Document-level template\n    paper_template: paper.md             # Paper-level template\n    paper_separator: \"\\n\\n---\\n\\n\"       # Separator between papers\n\n  # HTML export configuration\n  html:\n    template_dir: template/html/scholar/\n    document_template: document.html\n    paper_template: paper.html\n\n# llm domain\nllm:\n  enabled: false                          # Enable/disable LLM features\n  provider: openai-compat                 # Provider type (currently only this)\n  api_key_env: LLM_API_KEY                # Environment variable for API key\n  timeout: 30                             # Request timeout (seconds)\n  target_lang: Simplified Chinese         # Target language instruction passed to LLM\n  temperature: 0.0                        # Sampling temperature (0.0 = deterministic)\n  max_tokens: 800                         # Maximum response tokens\n  max_workers: 3                          # Parallel translation workers\n\n  # Feature selection\n  # You can choose one of three modes:\n  # 1. Translation only: enable_translation=true, enable_summary=false\n  # 2. Summary only: enable_translation=false, enable_summary=true\n  # 3. Both: enable_translation=true, enable_summary=true\n  enable_translation: false                # Enable abstract translation\n  enable_summary: false                   # Enable paper summary (TLDR, motivation, method, result, conclusion)\n                                          # Default: false (to control API costs)\n\n  # Retry configuration (for handling timeouts and transient errors)\n  max_retries: 3                          # Maximum retry attempts (0 to disable)\n  retry_base_delay: 1.0                   # Base delay for exponential backoff (seconds)\n  retry_max_delay: 10.0                   # Maximum delay between retries (seconds)\n  retry_timeout_multiplier: 1.0           # Timeout multiplier per retry (1.0 = no change)\n\n  # base_url and model are required when llm.enabled is true.\n  # Provider-specific configuration examples:\n  #\n  # OpenAI GPT-4o-mini (recommended for cost):\n  #   base_url: https://api.openai.com\n  #   model: gpt-4o-mini\n  #\n  # DeepSeek (very cost-effective):\n  #   base_url: https://api.deepseek.com\n  #   model: deepseek-chat\n  #\n  # SiliconFlow (free tier available):\n  #   base_url: https://api.siliconflow.cn\n  #   model: Qwen/Qwen2.5-7B-Instruct\n"}, "c7451fd7f8d0f0d6ffa7e8a677cab8cc47fc9900": {"t": 1785762184.9423068, "n": 4758, "h": "c67ab1b397b13d02", "c": "# Internal defaults.\n#\n# DO NOT edit this file to configure your instance.\n# To override any value, add the corresponding key to your own config file\n# (passed via --config). It will be deep-merged on top of these defaults.\n#\n# Required fields with no defaults — must be specified in user config:\n#   queries          — define at least one search query\n#   llm.base_url     — required when llm.enabled is true\n#   llm.model        — required when llm.enabled is true\n\n# runtime domain\nlog:\n  level: INFO\n  to_file: true\n  dir: log\n\n# dashboard domain\ndashboard:\n  host: 127.0.0.1\n  port: 8765\n  auto_refresh_seconds: 30\n\n# storage domain\nstorage:\n  enabled: true\n  db_path: database/papers.db     # Relative path (from working directory) or absolute path (starts with /)\n  content_storage_enabled: true   # save full paper content (default: true)\n  keep_arxiv_version: false       # false: 2601.21922, true: 2601.21922v1\n\n# Global scope (optional), applied to every query.\n# Field keys must be uppercase: TITLE/ABSTRACT/AUTHOR/JOURNAL/CATEGORY\n# Operator keys must be uppercase: AND/OR/NOT\n#\n# scope:\n#   CATEGORY:\n#     OR: [cs.CV]\n\nsearch:\n  sources: [arxiv]           # enabled search sources\n  max_results: 5              # max paper count for each query\n  ccf_enabled: false          # enable local CCF A/B whitelist helpers for supported sources\n  ccf_cache_path: database/ccf_venues_ab.json\n  ccf_ranks: [A, B]\n  dblp_recent_years: 2\n  openreview_recent_years: 2\n  openreview_max_pages: 3\n\n  # multi round fetch\n  pull_every: 7               # Strict time window, unit: day\n  fill_enabled: true          # Allow fill with paper outside the strict time window\n  max_lookback_days: 30       # Max length for fill window, -1 as infinite\n  max_fetch_items: 125        # Max number of items to fetch, -1 means unlimited\n  fetch_batch_size: 25        # Fetch count for each round\n  arxiv_min_interval_seconds: 12.0                  # Minimum spacing between arXiv requests across pages/retries\n  openalex_relevance_threshold: 0.0                    # local threshold for OpenAlex relevance_score\n\n# output domain\noutput:\n  base_dir: output/                      # Output root directory\n  formats: [console, json, html]         # Output formats: console, json, markdown, html (can select multiple)\n\n  # Markdown export configuration\n  markdown:\n    template_dir: template/markdown/\n    document_template: document.md       # Document-level template\n    paper_template: paper.md             # Paper-level template\n    paper_separator: \"\\n\\n---\\n\\n\"       # Separator between papers\n\n  # HTML export configuration\n  html:\n    template_dir: template/html/scholar/\n    document_template: document.html\n    paper_template: paper.html\n\n# llm domain\nllm:\n  enabled: false                          # Enable/disable LLM features\n  provider: openai-compat                 # Provider type (currently only this)\n  api_key_env: LLM_API_KEY                # Environment variable for API key\n  timeout: 30                             # Request timeout (seconds)\n  target_lang: Simplified Chinese         # Target language instruction passed to LLM\n  temperature: 0.0                        # Sampling temperature (0.0 = deterministic)\n  max_tokens: 800                         # Maximum response tokens\n  max_workers: 3                          # Parallel translation workers\n\n  # Feature selection\n  # You can choose one of three modes:\n  # 1. Translation only: enable_translation=true, enable_summary=false\n  # 2. Summary only: enable_translation=false, enable_summary=true\n  # 3. Both: enable_translation=true, enable_summary=true\n  enable_translation: false                # Enable abstract translation\n  enable_summary: false                   # Enable paper summary (TLDR, motivation, method, result, conclusion)\n                                          # Default: false (to control API costs)\n\n  # Retry configuration (for handling timeouts and transient errors)\n  max_retries: 3                          # Maximum retry attempts (0 to disable)\n  retry_base_delay: 1.0                   # Base delay for exponential backoff (seconds)\n  retry_max_delay: 10.0                   # Maximum delay between retries (seconds)\n  retry_timeout_multiplier: 1.0           # Timeout multiplier per retry (1.0 = no change)\n\n  # base_url and model are required when llm.enabled is true.\n  # Provider-specific configuration examples:\n  #\n  # OpenAI GPT-4o-mini (recommended for cost):\n  #   base_url: https://api.openai.com\n  #   model: gpt-4o-mini\n  #\n  # DeepSeek (very cost-effective):\n  #   base_url: https://api.deepseek.com\n  #   model: deepseek-chat\n  #\n  # SiliconFlow (free tier available):\n  #   base_url: https://api.siliconflow.cn\n  #   model: Qwen/Qwen2.5-7B-Instruct\n"}, "3159a02863847d1e183894af9461a57a94365eac": {"t": 1785762186.3438466, "n": 5025, "h": "75e499ba38be3772", "c": "# Internal defaults.\n#\n# DO NOT edit this file to configure your instance.\n# To override any value, add the corresponding key to your own config file\n# (passed via --config). It will be deep-merged on top of these defaults.\n#\n# Required fields with no defaults — must be specified in user config:\n#   queries          — define at least one search query\n#   llm.base_url     — required when llm.enabled is true\n#   llm.model        — required when llm.enabled is true\n\n# runtime domain\nlog:\n  level: INFO\n  to_file: true\n  dir: log\n\n# dashboard domain\ndashboard:\n  host: 127.0.0.1\n  port: 8765\n  auto_refresh_seconds: 30\n\n# storage domain\nstorage:\n  enabled: true\n  db_path: database/papers.db     # Relative path (from working directory) or absolute path (starts with /)\n  content_storage_enabled: true   # save full paper content (default: true)\n  keep_arxiv_version: false       # false: 2601.21922, true: 2601.21922v1\n\n# Global scope (optional), applied to every query.\n# Field keys must be uppercase: TITLE/ABSTRACT/AUTHOR/JOURNAL/CATEGORY\n# Operator keys must be uppercase: AND/OR/NOT\n#\n# scope:\n#   CATEGORY:\n#     OR: [cs.CV]\n\nsearch:\n  sources: [arxiv]           # enabled search sources\n  max_results: 5              # max paper count for each query\n  ccf_enabled: false          # enable local CCF A/B whitelist helpers for supported sources\n  ccf_cache_path: database/ccf_venues_ab.json\n  ccf_ranks: [A, B]\n  dblp_recent_years: 2\n  openreview_recent_years: 2\n  openreview_max_pages: 3\n\n  # multi round fetch\n  pull_every: 7               # Strict time window, unit: day\n  fill_enabled: true          # Allow fill with paper outside the strict time window\n  max_lookback_days: 30       # Max length for fill window, -1 as infinite\n  max_fetch_items: 125        # Max number of items to fetch, -1 means unlimited\n  fetch_batch_size: 25        # Fetch count for each round\n  arxiv_min_interval_seconds: 12.0                  # Minimum spacing between arXiv requests across pages/retries\n  openalex_relevance_threshold: 0.0                    # local threshold for OpenAlex relevance_score\n  ncbi_api_key_env: NCBI_API_KEY   # env var name for NCBI API key (optional, improves rate limit)\n  ncbi_tool: paper-tracker         # tool identifier sent to NCBI E-utilities\n  ncbi_email: \"\"                   # contact email sent to NCBI E-utilities (recommended)\n\n# output domain\noutput:\n  base_dir: output/                      # Output root directory\n  formats: [console, json, html]         # Output formats: console, json, markdown, html (can select multiple)\n\n  # Markdown export configuration\n  markdown:\n    template_dir: template/markdown/\n    document_template: document.md       # Document-level template\n    paper_template: paper.md             # Paper-level template\n    paper_separator: \"\\n\\n---\\n\\n\"       # Separator between papers\n\n  # HTML export configuration\n  html:\n    template_dir: template/html/scholar/\n    document_template: document.html\n    paper_template: paper.html\n\n# llm domain\nllm:\n  enabled: false                          # Enable/disable LLM features\n  provider: openai-compat                 # Provider type (currently only this)\n  api_key_env: LLM_API_KEY                # Environment variable for API key\n  timeout: 30                             # Request timeout (seconds)\n  target_lang: Simplified Chinese         # Target language instruction passed to LLM\n  temperature: 0.0                        # Sampling temperature (0.0 = deterministic)\n  max_tokens: 800                         # Maximum response tokens\n  max_workers: 3                          # Parallel translation workers\n\n  # Feature selection\n  # You can choose one of three modes:\n  # 1. Translation only: enable_translation=true, enable_summary=false\n  # 2. Summary only: enable_translation=false, enable_summary=true\n  # 3. Both: enable_translation=true, enable_summary=true\n  enable_translation: false                # Enable abstract translation\n  enable_summary: false                   # Enable paper summary (TLDR, motivation, method, result, conclusion)\n                                          # Default: false (to control API costs)\n\n  # Retry configuration (for handling timeouts and transient errors)\n  max_retries: 3                          # Maximum retry attempts (0 to disable)\n  retry_base_delay: 1.0                   # Base delay for exponential backoff (seconds)\n  retry_max_delay: 10.0                   # Maximum delay between retries (seconds)\n  retry_timeout_multiplier: 1.0           # Timeout multiplier per retry (1.0 = no change)\n\n  # base_url and model are required when llm.enabled is true.\n  # Provider-specific configuration examples:\n  #\n  # OpenAI GPT-4o-mini (recommended for cost):\n  #   base_url: https://api.openai.com\n  #   model: gpt-4o-mini\n  #\n  # DeepSeek (very cost-effective):\n  #   base_url: https://api.deepseek.com\n  #   model: deepseek-chat\n  #\n  # SiliconFlow (free tier available):\n  #   base_url: https://api.siliconflow.cn\n  #   model: Qwen/Qwen2.5-7B-Instruct\n"}, "89ef700fbca8d146e46079b0be419028ea491f82": {"t": 1785762188.6172755, "n": 5025, "h": "7c321846bf8d66b1", "c": "# Internal defaults.\n#\n# DO NOT edit this file to configure your instance.\n# To override any value, add the corresponding key to your own config file\n# (passed via --config). It will be deep-merged on top of these defaults.\n#\n# Required fields with no defaults — must be specified in user config:\n#   queries          — define at least one search query\n#   llm.base_url     — required when llm.enabled is true\n#   llm.model        — required when llm.enabled is true\n\n# runtime domain\nlog:\n  level: INFO\n  to_file: true\n  dir: log\n\n# dashboard domain\ndashboard:\n  host: 127.0.0.1\n  port: 8765\n  auto_refresh_seconds: 30\n\n# storage domain\nstorage:\n  enabled: true\n  db_path: database/papers.db     # Relative path (from working directory) or absolute path (starts with /)\n  content_storage_enabled: true   # save full paper content (default: true)\n  keep_arxiv_version: false       # false: 2601.21922, true: 2601.21922v1\n\n# Global scope (optional), applied to every query.\n# Field keys must be uppercase: TITLE/ABSTRACT/AUTHOR/JOURNAL/CATEGORY\n# Operator keys must be uppercase: AND/OR/NOT\n#\n# scope:\n#   CATEGORY:\n#     OR: [cs.CV]\n\nsearch:\n  sources: [arxiv]           # enabled search sources\n  max_results: 5              # max paper count for each query\n  ccf_enabled: false          # enable local CCF A/B whitelist helpers for supported sources\n  ccf_cache_path: database/ccf_venues_ab.json\n  ccf_ranks: [A, B]\n  dblp_recent_years: 2\n  openreview_recent_years: 2\n  openreview_max_pages: 3\n\n  # multi round fetch\n  pull_every: 7               # Strict time window, unit: day\n  fill_enabled: true          # Allow fill with paper outside the strict time window\n  max_lookback_days: 30       # Max length for fill window, -1 as infinite\n  max_fetch_items: 125        # Max number of items to fetch, -1 means unlimited\n  fetch_batch_size: 25        # Fetch count for each round\n  arxiv_min_interval_seconds: 5.0                   # Minimum spacing between arXiv requests across pages/retries\n  openalex_relevance_threshold: 0.0                    # local threshold for OpenAlex relevance_score\n  ncbi_api_key_env: NCBI_API_KEY   # env var name for NCBI API key (optional, improves rate limit)\n  ncbi_tool: paper-tracker         # tool identifier sent to NCBI E-utilities\n  ncbi_email: \"\"                   # contact email sent to NCBI E-utilities (recommended)\n\n# output domain\noutput:\n  base_dir: output/                      # Output root directory\n  formats: [console, json, html]         # Output formats: console, json, markdown, html (can select multiple)\n\n  # Markdown export configuration\n  markdown:\n    template_dir: template/markdown/\n    document_template: document.md       # Document-level template\n    paper_template: paper.md             # Paper-level template\n    paper_separator: \"\\n\\n---\\n\\n\"       # Separator between papers\n\n  # HTML export configuration\n  html:\n    template_dir: template/html/scholar/\n    document_template: document.html\n    paper_template: paper.html\n\n# llm domain\nllm:\n  enabled: false                          # Enable/disable LLM features\n  provider: openai-compat                 # Provider type (currently only this)\n  api_key_env: LLM_API_KEY                # Environment variable for API key\n  timeout: 30                             # Request timeout (seconds)\n  target_lang: Simplified Chinese         # Target language instruction passed to LLM\n  temperature: 0.0                        # Sampling temperature (0.0 = deterministic)\n  max_tokens: 800                         # Maximum response tokens\n  max_workers: 3                          # Parallel translation workers\n\n  # Feature selection\n  # You can choose one of three modes:\n  # 1. Translation only: enable_translation=true, enable_summary=false\n  # 2. Summary only: enable_translation=false, enable_summary=true\n  # 3. Both: enable_translation=true, enable_summary=true\n  enable_translation: false                # Enable abstract translation\n  enable_summary: false                   # Enable paper summary (TLDR, motivation, method, result, conclusion)\n                                          # Default: false (to control API costs)\n\n  # Retry configuration (for handling timeouts and transient errors)\n  max_retries: 3                          # Maximum retry attempts (0 to disable)\n  retry_base_delay: 1.0                   # Base delay for exponential backoff (seconds)\n  retry_max_delay: 10.0                   # Maximum delay between retries (seconds)\n  retry_timeout_multiplier: 1.0           # Timeout multiplier per retry (1.0 = no change)\n\n  # base_url and model are required when llm.enabled is true.\n  # Provider-specific configuration examples:\n  #\n  # OpenAI GPT-4o-mini (recommended for cost):\n  #   base_url: https://api.openai.com\n  #   model: gpt-4o-mini\n  #\n  # DeepSeek (very cost-effective):\n  #   base_url: https://api.deepseek.com\n  #   model: deepseek-chat\n  #\n  # SiliconFlow (free tier available):\n  #   base_url: https://api.siliconflow.cn\n  #   model: Qwen/Qwen2.5-7B-Instruct\n"}, "59615e9b21f240b547538c7d5b98c96d2b5194e0": {"t": 1785762189.9495847, "n": 4585, "h": "ddb43fc08c221879", "c": "# Internal defaults.\n#\n# DO NOT edit this file to configure your instance.\n# To override any value, add the corresponding key to your own config file\n# (passed via --config). It will be deep-merged on top of these defaults.\n#\n# Required fields with no defaults — must be specified in user config:\n#   queries          — define at least one search query\n#   llm.base_url     — required when llm.enabled is true\n#   llm.model        — required when llm.enabled is true\n\n# runtime domain\nlog:\n  level: INFO\n  to_file: true\n  dir: log\n\n# storage domain\nstorage:\n  enabled: true\n  db_path: database/papers.db     # Relative path (from working directory) or absolute path (starts with /)\n  content_storage_enabled: true   # save full paper content (default: true)\n  keep_arxiv_version: false       # false: 2601.21922, true: 2601.21922v1\n\n# Global scope (optional), applied to every query.\n# Field keys must be uppercase: TITLE/ABSTRACT/AUTHOR/JOURNAL/CATEGORY\n# Operator keys must be uppercase: AND/OR/NOT\n#\n# scope:\n#   CATEGORY:\n#     OR: [cs.CV]\n\nsearch:\n  sources: [arxiv]           # enabled search sources\n  max_results: 5              # max paper count for each query\n\n  # multi round fetch\n  pull_every: 7               # Strict time window, unit: day\n  fill_enabled: true          # Allow fill with paper outside the strict time window\n  max_lookback_days: 30       # Max length for fill window, -1 as infinite\n  max_fetch_items: 125        # Max number of items to fetch, -1 means unlimited\n  fetch_batch_size: 25        # Fetch count for each round\n  openalex_relevance_threshold: 0.0                    # local threshold for OpenAlex relevance_score\n  ncbi_api_key_env: NCBI_API_KEY   # env var name for NCBI API key (optional, improves rate limit)\n  ncbi_tool: paper-tracker         # tool identifier sent to NCBI E-utilities\n  ncbi_email: \"\"                   # contact email sent to NCBI E-utilities (recommended)\n\n# output domain\noutput:\n  base_dir: output/                      # Output root directory\n  formats: [console, json, html]         # Output formats: console, json, markdown, html (can select multiple)\n\n  # Markdown export configuration\n  markdown:\n    template_dir: template/markdown/\n    document_template: document.md       # Document-level template\n    paper_template: paper.md             # Paper-level template\n    paper_separator: \"\\n\\n---\\n\\n\"       # Separator between papers\n\n  # HTML export configuration\n  html:\n    template_dir: template/html/scholar/\n    document_template: document.html\n    paper_template: paper.html\n\n# llm domain\nllm:\n  enabled: false                          # Enable/disable LLM features\n  provider: openai-compat                 # Provider type (currently only this)\n  api_key_env: LLM_API_KEY                # Environment variable for API key\n  timeout: 30                             # Request timeout (seconds)\n  target_lang: Simplified Chinese         # Target language instruction passed to LLM\n  temperature: 0.0                        # Sampling temperature (0.0 = deterministic)\n  max_tokens: 800                         # Maximum response tokens\n  max_workers: 3                          # Parallel translation workers\n\n  # Feature selection\n  # You can choose one of three modes:\n  # 1. Translation only: enable_translation=true, enable_summary=false\n  # 2. Summary only: enable_translation=false, enable_summary=true\n  # 3. Both: enable_translation=true, enable_summary=true\n  enable_translation: false                # Enable abstract translation\n  enable_summary: false                   # Enable paper summary (TLDR, motivation, method, result, conclusion)\n                                          # Default: false (to control API costs)\n\n  # Retry configuration (for handling timeouts and transient errors)\n  max_retries: 3                          # Maximum retry attempts (0 to disable)\n  retry_base_delay: 1.0                   # Base delay for exponential backoff (seconds)\n  retry_max_delay: 10.0                   # Maximum delay between retries (seconds)\n  retry_timeout_multiplier: 1.0           # Timeout multiplier per retry (1.0 = no change)\n\n  # base_url and model are required when llm.enabled is true.\n  # Provider-specific configuration examples:\n  #\n  # OpenAI GPT-4o-mini (recommended for cost):\n  #   base_url: https://api.openai.com\n  #   model: gpt-4o-mini\n  #\n  # DeepSeek (very cost-effective):\n  #   base_url: https://api.deepseek.com\n  #   model: deepseek-chat\n  #\n  # SiliconFlow (free tier available):\n  #   base_url: https://api.siliconflow.cn\n  #   model: Qwen/Qwen2.5-7B-Instruct\n"}, "0584d250ad0d8b7dc1a9a7cc2b80cc65ca15c35b": {"t": 1785762191.612795, "n": 4318, "h": "25faf7d5e2717fd2", "c": "# Internal defaults.\n#\n# DO NOT edit this file to configure your instance.\n# To override any value, add the corresponding key to your own config file\n# (passed via --config). It will be deep-merged on top of these defaults.\n#\n# Required fields with no defaults — must be specified in user config:\n#   queries          — define at least one search query\n#   llm.base_url     — required when llm.enabled is true\n#   llm.model        — required when llm.enabled is true\n\n# runtime domain\nlog:\n  level: INFO\n  to_file: true\n  dir: log\n\n# storage domain\nstorage:\n  enabled: true\n  db_path: database/papers.db     # Relative path (from working directory) or absolute path (starts with /)\n  content_storage_enabled: true   # save full paper content (default: true)\n  keep_arxiv_version: false       # false: 2601.21922, true: 2601.21922v1\n\n# Global scope (optional), applied to every query.\n# Field keys must be uppercase: TITLE/ABSTRACT/AUTHOR/JOURNAL/CATEGORY\n# Operator keys must be uppercase: AND/OR/NOT\n#\n# scope:\n#   CATEGORY:\n#     OR: [cs.CV]\n\nsearch:\n  sources: [arxiv]           # enabled search sources\n  max_results: 5              # max paper count for each query\n\n  # multi round fetch\n  pull_every: 7               # Strict time window, unit: day\n  fill_enabled: true          # Allow fill with paper outside the strict time window\n  max_lookback_days: 30       # Max length for fill window, -1 as infinite\n  max_fetch_items: 125        # Max number of items to fetch, -1 means unlimited\n  fetch_batch_size: 25        # Fetch count for each round\n  openalex_relevance_threshold: 0.0                    # local threshold for OpenAlex relevance_score\n\n# output domain\noutput:\n  base_dir: output/                      # Output root directory\n  formats: [console, json, html]         # Output formats: console, json, markdown, html (can select multiple)\n\n  # Markdown export configuration\n  markdown:\n    template_dir: template/markdown/\n    document_template: document.md       # Document-level template\n    paper_template: paper.md             # Paper-level template\n    paper_separator: \"\\n\\n---\\n\\n\"       # Separator between papers\n\n  # HTML export configuration\n  html:\n    template_dir: template/html/scholar/\n    document_template: document.html\n    paper_template: paper.html\n\n# llm domain\nllm:\n  enabled: false                          # Enable/disable LLM features\n  provider: openai-compat                 # Provider type (currently only this)\n  api_key_env: LLM_API_KEY                # Environment variable for API key\n  timeout: 30                             # Request timeout (seconds)\n  target_lang: Simplified Chinese         # Target language instruction passed to LLM\n  temperature: 0.0                        # Sampling temperature (0.0 = deterministic)\n  max_tokens: 800                         # Maximum response tokens\n  max_workers: 3                          # Parallel translation workers\n\n  # Feature selection\n  # You can choose one of three modes:\n  # 1. Translation only: enable_translation=true, enable_summary=false\n  # 2. Summary only: enable_translation=false, enable_summary=true\n  # 3. Both: enable_translation=true, enable_summary=true\n  enable_translation: false                # Enable abstract translation\n  enable_summary: false                   # Enable paper summary (TLDR, motivation, method, result, conclusion)\n                                          # Default: false (to control API costs)\n\n  # Retry configuration (for handling timeouts and transient errors)\n  max_retries: 3                          # Maximum retry attempts (0 to disable)\n  retry_base_delay: 1.0                   # Base delay for exponential backoff (seconds)\n  retry_max_delay: 10.0                   # Maximum delay between retries (seconds)\n  retry_timeout_multiplier: 1.0           # Timeout multiplier per retry (1.0 = no change)\n\n  # base_url and model are required when llm.enabled is true.\n  # Provider-specific configuration examples:\n  #\n  # OpenAI GPT-4o-mini (recommended for cost):\n  #   base_url: https://api.openai.com\n  #   model: gpt-4o-mini\n  #\n  # DeepSeek (very cost-effective):\n  #   base_url: https://api.deepseek.com\n  #   model: deepseek-chat\n  #\n  # SiliconFlow (free tier available):\n  #   base_url: https://api.siliconflow.cn\n  #   model: Qwen/Qwen2.5-7B-Instruct\n"}, "22bec1e9c8f86e7a51f4b339be4e6c6d86075ba7": {"t": 1785762194.127001, "n": 4216, "h": "d7c571e65dcf0937", "c": "# Internal defaults.\n#\n# DO NOT edit this file to configure your instance.\n# To override any value, add the corresponding key to your own config file\n# (passed via --config). It will be deep-merged on top of these defaults.\n#\n# Required fields with no defaults — must be specified in user config:\n#   queries          — define at least one search query\n#   llm.base_url     — required when llm.enabled is true\n#   llm.model        — required when llm.enabled is true\n\n# runtime domain\nlog:\n  level: INFO\n  to_file: true\n  dir: log\n\n# storage domain\nstorage:\n  enabled: true\n  db_path: database/papers.db     # Relative path (from working directory) or absolute path (starts with /)\n  content_storage_enabled: true   # save full paper content (default: true)\n  keep_arxiv_version: false       # false: 2601.21922, true: 2601.21922v1\n\n# Global scope (optional), applied to every query.\n# Field keys must be uppercase: TITLE/ABSTRACT/AUTHOR/JOURNAL/CATEGORY\n# Operator keys must be uppercase: AND/OR/NOT\n#\n# scope:\n#   CATEGORY:\n#     OR: [cs.CV]\n\nsearch:\n  sources: [arxiv]           # enabled search sources\n  max_results: 5              # max paper count for each query\n\n  # multi round fetch\n  pull_every: 7               # Strict time window, unit: day\n  fill_enabled: true          # Allow fill with paper outside the strict time window\n  max_lookback_days: 30       # Max length for fill window, -1 as infinite\n  max_fetch_items: 125        # Max number of items to fetch, -1 means unlimited\n  fetch_batch_size: 25        # Fetch count for each round\n\n# output domain\noutput:\n  base_dir: output/                      # Output root directory\n  formats: [console, json, html]         # Output formats: console, json, markdown, html (can select multiple)\n\n  # Markdown export configuration\n  markdown:\n    template_dir: template/markdown/\n    document_template: document.md       # Document-level template\n    paper_template: paper.md             # Paper-level template\n    paper_separator: \"\\n\\n---\\n\\n\"       # Separator between papers\n\n  # HTML export configuration\n  html:\n    template_dir: template/html/scholar/\n    document_template: document.html\n    paper_template: paper.html\n\n# llm domain\nllm:\n  enabled: false                          # Enable/disable LLM features\n  provider: openai-compat                 # Provider type (currently only this)\n  api_key_env: LLM_API_KEY                # Environment variable for API key\n  timeout: 30                             # Request timeout (seconds)\n  target_lang: Simplified Chinese         # Target language instruction passed to LLM\n  temperature: 0.0                        # Sampling temperature (0.0 = deterministic)\n  max_tokens: 800                         # Maximum response tokens\n  max_workers: 3                          # Parallel translation workers\n\n  # Feature selection\n  # You can choose one of three modes:\n  # 1. Translation only: enable_translation=true, enable_summary=false\n  # 2. Summary only: enable_translation=false, enable_summary=true\n  # 3. Both: enable_translation=true, enable_summary=true\n  enable_translation: false                # Enable abstract translation\n  enable_summary: false                   # Enable paper summary (TLDR, motivation, method, result, conclusion)\n                                          # Default: false (to control API costs)\n\n  # Retry configuration (for handling timeouts and transient errors)\n  max_retries: 3                          # Maximum retry attempts (0 to disable)\n  retry_base_delay: 1.0                   # Base delay for exponential backoff (seconds)\n  retry_max_delay: 10.0                   # Maximum delay between retries (seconds)\n  retry_timeout_multiplier: 1.0           # Timeout multiplier per retry (1.0 = no change)\n\n  # base_url and model are required when llm.enabled is true.\n  # Provider-specific configuration examples:\n  #\n  # OpenAI GPT-4o-mini (recommended for cost):\n  #   base_url: https://api.openai.com\n  #   model: gpt-4o-mini\n  #\n  # DeepSeek (very cost-effective):\n  #   base_url: https://api.deepseek.com\n  #   model: deepseek-chat\n  #\n  # SiliconFlow (free tier available):\n  #   base_url: https://api.siliconflow.cn\n  #   model: Qwen/Qwen2.5-7B-Instruct\n"}, "7370702a096221c4ee62c8650969af2297ef9fb0": {"t": 1785762200.797297, "n": 4162, "h": "6983069516189a91", "c": "# Internal defaults.\n#\n# DO NOT edit this file to configure your instance.\n# To override any value, add the corresponding key to your own config file\n# (passed via --config). It will be deep-merged on top of these defaults.\n#\n# Required fields with no defaults — must be specified in user config:\n#   queries          — define at least one search query\n#   llm.base_url     — required when llm.enabled is true\n#   llm.model        — required when llm.enabled is true\n\n# runtime domain\nlog:\n  level: INFO\n  to_file: true\n  dir: log\n\n# storage domain\nstorage:\n  enabled: true\n  db_path: database/papers.db     # Relative path (from working directory) or absolute path (starts with /)\n  content_storage_enabled: true   # save full paper content (default: true)\n  keep_arxiv_version: false       # false: 2601.21922, true: 2601.21922v1\n\n# Global scope (optional), applied to every query.\n# Field keys must be uppercase: TITLE/ABSTRACT/AUTHOR/JOURNAL/CATEGORY\n# Operator keys must be uppercase: AND/OR/NOT\n#\n# scope:\n#   CATEGORY:\n#     OR: [cs.CV]\n\nsearch:\n  max_results: 5              # max paper count for each query\n\n  # multi round fetch\n  pull_every: 7               # Strict time window, unit: day\n  fill_enabled: true          # Allow fill with paper outside the strict time window\n  max_lookback_days: 30       # Max length for fill window, -1 as infinite\n  max_fetch_items: 125        # Max number of items to fetch, -1 means unlimited\n  fetch_batch_size: 25        # Fetch count for each round\n\n# output domain\noutput:\n  base_dir: output/                      # Output root directory\n  formats: [console, json, html]         # Output formats: console, json, markdown, html (can select multiple)\n\n  # Markdown export configuration\n  markdown:\n    template_dir: template/markdown/\n    document_template: document.md       # Document-level template\n    paper_template: paper.md             # Paper-level template\n    paper_separator: \"\\n\\n---\\n\\n\"       # Separator between papers\n\n  # HTML export configuration\n  html:\n    template_dir: template/html/scholar/\n    document_template: document.html\n    paper_template: paper.html\n\n# llm domain\nllm:\n  enabled: false                          # Enable/disable LLM features\n  provider: openai-compat                 # Provider type (currently only this)\n  api_key_env: LLM_API_KEY                # Environment variable for API key\n  timeout: 30                             # Request timeout (seconds)\n  target_lang: Simplified Chinese         # Target language instruction passed to LLM\n  temperature: 0.0                        # Sampling temperature (0.0 = deterministic)\n  max_tokens: 800                         # Maximum response tokens\n  max_workers: 3                          # Parallel translation workers\n\n  # Feature selection\n  # You can choose one of three modes:\n  # 1. Translation only: enable_translation=true, enable_summary=false\n  # 2. Summary only: enable_translation=false, enable_summary=true\n  # 3. Both: enable_translation=true, enable_summary=true\n  enable_translation: false                # Enable abstract translation\n  enable_summary: false                   # Enable paper summary (TLDR, motivation, method, result, conclusion)\n                                          # Default: false (to control API costs)\n\n  # Retry configuration (for handling timeouts and transient errors)\n  max_retries: 3                          # Maximum retry attempts (0 to disable)\n  retry_base_delay: 1.0                   # Base delay for exponential backoff (seconds)\n  retry_max_delay: 10.0                   # Maximum delay between retries (seconds)\n  retry_timeout_multiplier: 1.0           # Timeout multiplier per retry (1.0 = no change)\n\n  # base_url and model are required when llm.enabled is true.\n  # Provider-specific configuration examples:\n  #\n  # OpenAI GPT-4o-mini (recommended for cost):\n  #   base_url: https://api.openai.com\n  #   model: gpt-4o-mini\n  #\n  # DeepSeek (very cost-effective):\n  #   base_url: https://api.deepseek.com\n  #   model: deepseek-chat\n  #\n  # SiliconFlow (free tier available):\n  #   base_url: https://api.siliconflow.cn\n  #   model: Qwen/Qwen2.5-7B-Instruct\n"}}}}