What problem does it solve? Autonomous wiki tooling drifts over time: lint warnings accumulate, scripts lack tests, skills bloat, and scheduled runs waste tokens producing nothing. This Skill provides a disciplined, measurement-driven process for improving the wiki's machinery — scripts, hooks, lint rules, skills, and settings — one verified fix at a time. ## Core Features & Use Cases - Six-Step Enforcement Loop: Runs MEASURE → IDENTIFY → FIX → ENFORCE → VERIFY → LOG on every invocation, requiring a sea health snapshot before and after each change. - Priority-Stack Issue Selection: Ranks candidate fixes from user frustration signals in transcripts (P0) down through lint errors, missing tests, workflow waste, and token bloat (P8), enforcing one problem per run. - Code-Level Enforcement: Pushes every fix down an enforcement stack — deny rules, hooks, lint rules, then skill edits — and requires a regression-prevention mechanism such as a test or registered hook. - Use Case: A scheduled daily routine fires, takes a health snapshot showing 143 deadend pages, adds a lint rule via TDD, verifies the count drops to 131, and commits feat: CSI — add wiki_lint rule (deadend_count: 143 → 131). ## Quick Start Ask the agent to run a continuous self-improvement pass on the wiki infrastructure and report the before-and-after health metrics.
