context-file-generation

To generate subdirectory CLAUDE.md files, run the structured-workflows setup operation, which scans targeted directories and applies content assembly rules to create provenance-marked context files for specific codebase areas.

The three-layer context model structures repository documentation into root, subdirectory, and skills/agents layers, ensuring AI agents receive domain-specific ambient context and local conventions for reliable operation within targeted codebase areas.

Migrating root documentation into subdirectory CLAUDE.md files uses root migrations and demoted-skill synthesis to preserve provenance markers, avoid redundant content, and generate workflow.yaml reconciliation entries for safe, auditable file tracking.

Yes, updating existing CLAUDE.md files uses interaction modes like lean-update and lean-append, applying provenance markers and corruption handling to safely modify content while preserving original context and maintaining workflow.yaml reconciliation entries.

Orphan detection tracks generated and migrated files during setup by cross-referencing workflow.yaml reconciliation entries, identifying unmanaged or displaced context files to ensure repeatable, auditable context generation across the repository structure.

Automated context generation requires qualifying subdirectories to scan and relies on demoted-skill synthesis and directory scans; it may not suit repositories lacking clear structural separation or those needing highly customized, non-standard ambient context formatting.