Get D4D HTML from new storage location #119
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| # D4D Assistant Deterministic Configuration | ||
|
Check failure on line 1 in .github/workflows/d4d_assistant_deterministic_config.yaml
|
||
| # Ensures reproducible D4D generation matching Claude Code agent approach | ||
| # | ||
| # Purpose: Centralized configuration for deterministic D4D extraction | ||
| # Version: 1.0.0 | ||
| # Last Updated: 2025-02-10 | ||
| # Model Configuration | ||
| model: | ||
| provider: anthropic | ||
| name: claude-sonnet-4-5-20250929 # Date-pinned for reproducibility | ||
| temperature: 0.0 # Maximum determinism | ||
| max_tokens: 16000 # Sufficient for comprehensive D4D generation | ||
| # Prompt Configuration | ||
| prompts: | ||
| # External version-controlled prompt files | ||
| system_prompt_file: src/download/prompts/d4d_concatenated_system_prompt.txt | ||
| user_prompt_file: src/download/prompts/d4d_concatenated_user_prompt.txt | ||
| # Prompt templates can be overridden for specific modes | ||
| file_mode_prompt: null # Future: src/download/prompts/d4d_assistant_file_mode_prompt.txt | ||
| url_mode_prompt: null # Future: src/download/prompts/d4d_assistant_url_mode_prompt.txt | ||
| # Schema Configuration | ||
| schema: | ||
| source: local # Use local version-controlled schema file | ||
| path: src/data_sheets_schema/schema/data_sheets_schema_all.yaml | ||
| # Fallback to remote if local not available (not recommended) | ||
| remote_url: https://raw.githubusercontent.com/bridge2ai/data-sheets-schema/main/src/data_sheets_schema/schema/data_sheets_schema_all.yaml | ||
| # Metadata Generation | ||
| metadata: | ||
| generate_metadata: true # Always generate comprehensive metadata | ||
| track_input_hash: true # SHA-256 hash of input files/URLs | ||
| track_schema_hash: true # SHA-256 hash of schema file | ||
| track_prompt_hashes: true # SHA-256 hashes of prompt files | ||
| track_git_commit: true # Current git commit for provenance | ||
| include_environment: true # Platform, Python version, etc. | ||
| # Input Modes | ||
| input: | ||
| # File-based mode (preferred for reproducibility) | ||
| file_mode: | ||
| enabled: true | ||
| input_dir: data/sheets_d4dassistant/inputs | ||
| supported_formats: [txt, md, json, yaml, pdf, html] | ||
| # URL-based mode (fallback when files not provided) | ||
| url_mode: | ||
| enabled: true | ||
| download_dir: data/sheets_d4dassistant/fetched | ||
| timeout_seconds: 30 | ||
| max_retries: 3 | ||
| # Output Configuration | ||
| output: | ||
| base_dir: data/sheets_d4dassistant | ||
| yaml_filename_pattern: "{dataset}_d4d.yaml" | ||
| metadata_filename_pattern: "{dataset}_d4d_metadata.yaml" | ||
| html_filename_pattern: "{dataset}_d4d.html" | ||
| # Validation Gates | ||
| validation: | ||
| # Quality thresholds (BLOCK PR if not met) | ||
| min_sections_comprehensive: 10 # All D4D sections populated | ||
| min_sections_acceptable: 7 # Most D4D sections populated | ||
| min_sections_minimal: 4 # Basic D4D sections | ||
| min_slots_comprehensive: 146 # Comprehensive datasheet | ||
| min_slots_acceptable: 100 # Acceptable quality | ||
| min_slots_minimal: 50 # Minimal viable | ||
| min_lines_comprehensive: 200 # File size indicator | ||
| min_lines_acceptable: 150 | ||
| min_lines_minimal: 100 | ||
| # Required fields (always enforced) | ||
| required_fields: [id, name] | ||
| # Block PR creation if below this threshold | ||
| block_threshold: minimal # Options: comprehensive, acceptable, minimal, none | ||
| # Run LinkML schema validation before PR | ||
| schema_validation: true | ||
| # Reproducibility Settings | ||
| reproducibility: | ||
| # Settings that ensure determinism | ||
| deterministic: true | ||
| random_seed: null # Not applicable for LLM calls at temp=0.0 | ||
| # Verification settings | ||
| verify_reproducibility: false # Future: Run twice and compare outputs | ||
| diff_tolerance: 0 # Exact match required | ||
| # Provenance Tracking | ||
| provenance: | ||
| extraction_type: github_assistant_claude_code | ||
| performed_by: d4dassistant | ||
| wrapper_script: src/github/generate_d4d_metadata.py | ||
| version: "1.0.0" | ||
| # Prerequisites Validation | ||
| prerequisites: | ||
| # Check these before D4D generation (FAIL FAST if missing) | ||
| check_schema_exists: true | ||
| check_prompts_exist: true | ||
| check_input_files: true # File mode | ||
| check_urls_accessible: true # URL mode | ||
| # Timeout for prerequisites checks | ||
| check_timeout_seconds: 10 | ||
| # Error Handling | ||
| error_handling: | ||
| # What to do on various failures | ||
| on_validation_failure: fail # Options: fail, warn, ignore | ||
| on_prerequisites_failure: fail # Options: fail, warn | ||
| on_schema_load_failure: fail # Options: fail, fallback_remote | ||
| # Retry settings | ||
| max_generation_attempts: 1 # No retries for deterministic generation | ||
| retry_on_validation_failure: false | ||
| # Budget and Cost Tracking (optional) | ||
| budget: | ||
| check_before_generation: false # Future: check CBORG API budget | ||
| warn_threshold_percent: 75 # Warn if > 75% of budget used | ||
| api_url: null # Future: CBORG API endpoint | ||
| # Documentation | ||
| docs: | ||
| readme_path: data/sheets_d4dassistant/README.md | ||
| create_instructions_path: .github/workflows/d4d_assistant_create.md | ||
| edit_instructions_path: .github/workflows/d4d_assistant_edit.md | ||
| # Feature Flags (for gradual rollout) | ||
| features: | ||
| enable_metadata_generation: true | ||
| enable_completeness_validation: true | ||
| enable_prerequisites_check: true | ||
| enable_file_mode: true | ||
| enable_url_mode: true | ||
| enable_batch_mode: false # Future: process multiple datasets | ||
