What problem does it solve? Codebases accumulate gaps between what documentation promises and what code delivers: untested public APIs, stale TODOs, missing sibling implementations, and operator-to-chart parity drift. This Skill finds those gaps and fills the safe ones in a single end-to-end pass, rather than only reporting them. ## Core Features & Use Cases - Eleven gap-source enumeration: Scans doc-vs-code promises, untested exported symbols, TODO/FIXME markers, stale issue references, missing error handling, convention drift, env-var claims, and operator-to-helm-chart parity. - Risk-tier gating (1-10): A tier ladder controls which gaps are eligible for auto-fill, from cosmetic cleanups at tier 1 to architectural changes at tier 9, with above-tier candidates flagged for later runs. - Validation gauntlet and per-gap commits: Each candidate passes a five-concern gauntlet and per-tier fix-bar before being filled, committed individually, and recorded in a findings memory file. - Use Case: A dispatcher sends "fill gaps in operator tier 5"; the Skill enumerates operator-section gaps, fills the ones at or below tier 5 with one commit each, flags the rest, and delegates push and CI watch to a peer agent. ## Quick Start Ask the agent to fill gaps in the operator section at tier 5 and report what was filled versus flagged.
