A working adoption: one policy per layer (hook, rule, skill).
the framework → · the full catalog → · just the bare scaffold →
Demo repository. Three policies instead of the full catalog's twenty-eight, so every moving part fits in one sitting. Questions and issues belong on the framework repo.
Click Use this template above to start your own adoption.
| Layer | Policy | Enforcement | Mechanism |
|---|---|---|---|
| Hook | protect-main-branch |
Enforced — blocks | A compiled gate in .chock/compiled/protect-main-branch/git-hook/, wired into .git/hooks: commits, merges and pushes to main fail (pre-commit, pre-merge-commit, pre-push, plus a CI gate step) |
| Rule | block-no-verify |
Advisory | Ambient text compiled into AGENTS.md and every agent wrapper, plus a compiled PreToolUse guard blocking --no-verify at tool time in Claude Code |
| Skill | commit-message-style |
Invoked, not gated | Runs when the agent's task matches; evals/suite.yaml defines what "working" means |
Hook. Try it: git checkout main && echo x >> README.md && git add . && git commit -m "direct to main" fails outright — Direct commits/pushes to a protected branch (main|master) are blocked. Create a feature branch and open a pull request. The commit
never lands.
Rule. An agent reading AGENTS.md sees the same instruction as ambient text before it
acts. In Claude Code specifically, a compiled PreToolUse guard also blocks the
--no-verify/-n flag at tool-call time, so skipping the hook isn't an option even if the
agent tries it.
Skill. commit-message-style doesn't block anything — it's invoked when the task
matches (writing a commit message), and evals/suite.yaml is the test of whether it did
that well.
Full detail per policy: .agents/policies/ and
.agents/skills/commit-message-style/. More on how
this repo was built, and where to go next: docs/README.md.
| agentseam | the primitives — one handler API and a verified capability matrix across 16 agents |
| chock | the compiler — one policy into git hooks, CI gates and native pre-tool hooks |
| chock-catalog | the policies — 39, each labelled enforced or advisory, with replayed evals |
| context-report | the evidence — a signed report of whether an agent artifact actually works |
| chock-threat-intel | the threat ledger the catalog's policies answer to |
| chock-{claude,cursor,copilot,codex}-plugins | the catalog, packaged for each agent's plugin format (generated) |
| chock-quickstart · chock-example | template repos: what chock init leaves behind, and a full adoption |
Apache-2.0 — see LICENSE.