Extract

Extract Codebase Knowledge

Build or rebuild the .gauntlet/knowledge.json knowledge base.

When NOT To Use

• Tribal knowledge no parser can see (use gauntlet:curate)
• Building the code graph (use gauntlet:graph-build)

Steps

• Identify target directory: use the current working directory

or a user-specified path

• Run AST extraction: invoke the extractor script

   python3 ${CLAUDE_PLUGIN_ROOT}/scripts/extractor.py <target-dir>

• AI enrichment: for each extracted entry, enhance the detail

field with natural language explanation of business logic, data

flow, architectural role, and rationale

• Cross-reference: link related entries across modules by

matching imports, shared types, and data flow paths

• Merge with annotations: preserve existing curated entries

in .gauntlet/annotations/

• Save: write to .gauntlet/knowledge.json

• Report: show summary by category, coverage gaps, difficulty

distribution

Exit Criteria

• [ ] .gauntlet/knowledge.json exists and is valid JSON after the

skill completes; entries from .gauntlet/annotations/ are merged

and not overwritten

• [ ] Report shows entry counts broken down by all 7 categories

(business_logic, architecture, data_flow, api_contract, pattern,

dependency, error_handling) with coverage gaps identified

• [ ] Each extracted entry has a detail field containing a natural

language explanation (not just the raw AST node name)

• [ ] Cross-reference links between related entries are present for

modules sharing imports, shared types, or data flow paths

Category Priority

• business_logic (weight 7)
• architecture (weight 6)
• data_flow (weight 5)
• api_contract (weight 4)
• pattern (weight 3)
• dependency (weight 2)
• error_handling (weight 1)