prune-harness

Automating documentation pruning applies signal-driven analysis across the repository to identify outdated rules and content. This generates a structured pruning proposal with rationales, writing an auditable candidate list directly to a planning markdown file.

Signal-driven analysis evaluates rules, references, and planning documents to surface deletion candidates. It applies multiple signals to scan target files and generates a formal pruning proposal with specific signal justifications for each identified candidate.

Generating an auditable pruning proposal scans all target files under the content directory and related docs. The workflow writes the candidate list, rationales, and signal justifications to a timestamped planning file without altering the original content.

No, the pruning analysis workflow never modifies original content files. It scans target files under the content directory and related docs, terminating with a complete proposal written to the planning directory even when no deletion candidates exist.

When no outdated rules are found, the analysis still terminates with a complete proposal. It writes an auditable candidate list to the planning directory, preserving the audit trail and documenting the signal-driven analysis results for that date.