Claude Code Kit

An unsupervised AI agent will eventually skip a safety check under pressure — and nothing alarms when it does.

Under deadline stress it force-pushes past a failing test, commits without updating docs, or brings back a pattern you deleted on purpose. These failures are silent: they accumulate until your codebase is inconsistent or a secret is sitting in git history. Better prompts don't fix this — rules in a chat window evaporate; rules enforced by the toolchain do not.

A drop-in governance layer that forces every commit through an AI reviewer and five automated quality gates — installed in one command.

You get three things a buyer cares about. Auditability: every commit is reviewed for style, risk, and secrets before it lands. Recoverability: hard hooks physically block force-push and history rewrites — the agent cannot bypass them even if told to. Consistency: docs freshness, link integrity, and markdown budgets are checked automatically, every time.

Three enforcement layers run on every commit, and the agent cannot opt out of any of them.

Claude Code hooks fire before any command and block 11 dangerous git patterns — force-push, reset --hard, history rewrites — at the system level. A two-phase pre-commit pipeline then runs an AI judge on each changed file, followed by five deterministic checks (script reasons, link integrity, docs freshness, markdown budgets, secret scanning). A short rules file is re-injected on every commit so a long session never erodes your constraints.

Unchanged files are never re-reviewed, and no API key is required to start.

The AI judge caches each verdict by content hash, so commit-hook time stays flat as your codebase grows. The judge is fully opt-in: if no key is set, the gate soft-passes and the commit proceeds — safe to adopt with zero cost. Everything installs under a namespaced folder, so it never collides with your existing tools, and your CLAUDE.md is never overwritten.

Born from a real project of 2,871 commits built over ~83 days with AI as the primary developer.

The kit was extracted and generalized from a single-developer codebase where every failure mode it guards against — force-push under pressure, doc drift, secret leak — actually happened and was fixed by this exact discipline. It installs with one command, runs on Python 3.9 and git, and ships with 9 reusable workflow skills (handoff, TDD, issue creation, documentation audits, and more).