Extract atomic features with source evidence into features_raw.json.
npx skills add https://github.com/Swear4/agent_rules --skill feature-miner-swear4
Please help me install this Agent Skill. Skill: feature-miner Source: https://github.com/Swear4/agent_rules/tree/main/agent-universal/skills/feature-miner Command: npx skills add https://github.com/Swear4/agent_rules --skill feature-miner-swear4
Feature Miner analyzes source code file by file to extract atomic functionalities with evidence. It uses a structured, triple-pass cognitive approach to ensure precise discovery after project-mapper.
Run the feature-miner after project-mapper has produced codebase_map.json and review the resulting features_raw.json.
High-intent search queries and answers about installing and using this skill.
Feature extraction identifies atomic functionalities from source code by attaching verifiable proof from real functions and files. It uses a structured triple-pass approach to generate a features_raw.json inventory of macro-systems, mid-level pépites, and micro-capacities for audit, refactoring, and documentation.
You generate a feature inventory by analyzing source code file by file to extract cohesive features with explicit sourceEvidence. This produces a features_raw.json file enforcing strict validation rules, ensuring every extracted feature is grounded in code for accurate refactoring and auditing.
Yes, you need a codebase_map.json produced by a project-mapper before extracting code features. Feature extraction operates after codebase_map.json is produced, using the mapped structure to generate the final features_raw.json inventory for roadmap planning and QA.
The best way to ensure traceability during code-analysis is to enforce strict validation where extracted features must be grounded in code and include explicit sourceEvidence. This guarantees every macro-system, mid-level pépite, and micro-capacity includes verifiable proof sourced from real functions and files.
Feature extraction works with any codebase structure as long as codebase_map.json has been produced first. It analyzes source files file by file to identify cohesive atomic features, attaching verifiable proofs from real functions regardless of the underlying project layout or architecture.
MINT-IACoordinate parallel agents to generate seven codebase documents in .planning/codebase/.
openaiNormalize extracted financial rows into a model-ready long-form schema with QA artifacts.
VoDaiLoczAutomate linting and static analysis of code in CI pipelines.
mehdicScan repository contents to detect patterns and extract reusable utilities.
