speckit-adlc-levelup-init

You can discover implicit patterns by scanning a brownfield codebase to surface them as draft Context Directive Records (CDRs). This skill analyzes project structure, classifies patterns into context types, and generates CDR drafts with supporting evidence for team-ai-directives.

A Context Directive Record (CDR) captures implicit coding patterns and project context to standardize team-ai-directives. The skill generates CDR drafts by analyzing brownfield codebases, deduplicating against existing directives to ensure consistent AI governance and codebase reuse.

To generate CDR drafts, run the levelup init workflow to scan your codebase. The skill analyzes project structure, applies heuristic controls via the --cdr-heuristic flag, and writes the resulting pattern classifications to the {REPO_ROOT}/.specify/drafts/cdr.md file.

Yes, generated CDR drafts support workflow integration through handoff options. After drafts are written to {REPO_ROOT}/.specify/drafts/cdr.md, you can use the clarify, specify, and implement steps to refine and deploy the directives into your active software engineering pipeline.

Yes, the CDR discovery tool deduplicates newly found patterns against existing team-ai-directives. It applies heuristic controls via --cdr-heuristic to classify patterns into context types and skill categories, ensuring only novel directives are drafted in the cdr.md file.

When scanning a brownfield codebase, the CDR discovery outputs drafts to {REPO_ROOT}/.specify/drafts/cdr.md rather than finalizing them. Users must manually review these drafts and utilize the clarify, specify, and implement handoff steps to finalize the directives for production.