lens-feature-yaml

Run the feature-yaml-ops.py script with the read subcommand, passing --feature-id or --feature-path plus --governance-repo. It returns JSON containing identity fields, phase, track, docs paths, target_repos, dependencies, milestones, and transition history.

Use the validate subcommand with --to-phase to check the target phase against the lifecycle contract and feature track. Invalid transitions return a fail payload listing allowed next phases, while missing target_repos on implementation-impacting tracks returns a non-blocking warning.

Yes, the update subcommand surgically modifies only requested fields: phase, docs.path, docs.governance_docs_path, target_repos, and milestones. Unknown and untouched fields remain in place, and writes are atomic via a temporary file replace.

The commit-dirty subcommand detects dirty state, pulls with rebase and autostash, stages requested paths, commits with a message, pushes, and reports the resulting SHA. If nothing relevant is dirty, it just pulls and reports no commit.

The script emits a JSON fail payload with the error code invalid_phase_transition, the current and target phases, and the list of allowed next phases. It exits with a non-zero code so callers can block the workflow.

Only a fixed set of fields can be updated: phase, docs paths, target_repos, milestones, and links.pull_request. Other fields are rejected as unsupported, and phase changes must still pass lifecycle transition validation.