loop-audit
01Skill Claude CodeCodex
Audit the loop against loop-engineering principles and decision-gate adherence; write a scored report to audit.md.
Claude Code plugin for loop engineering: implement-verify cycles with machine-verifiable stop conditions, an independent read-only verifier agent, and disk-based state that survives session death
Skill Claude CodeCodex
Audit the loop against loop-engineering principles and decision-gate adherence; write a scored report to audit.md.
Skill Claude CodeCodex
Scaffold a GitHub Actions CI workflow from the loop's detected test/lint/build commands.
Skill Claude CodeCodex
Loop engineering method — machine-verifiable stop conditions, maker/checker separation (Codex CLI or Claude as maker, a verifier subagent as checker), disk-based memory in .claude/loop/. Consult when designing or running agent loops.
Skill Claude CodeCodex
Scaffold .claude/loop/ state files and detect this project's test/lint/build commands.
Skill Claude CodeCodex
Comprehensively review a project's loop/harness engineering against the control plane — ETCLOVG coverage, maturity, adversarial red-team — writing a scored harness-review.md; with --fix, remediate the high-severity findings on a branch via loop-run and stop at a PR.
Skill Claude CodeCodex
Run maker/checker cycles — implement via Codex CLI or Claude per loop.config.md, verifier grades — until the rubric passes; supports --once and --verify-only.
Skill Claude CodeCodex
Summarize loop state, rubric progress, and the last run token estimate from .claude/loop/.
Skill Claude CodeCodex
Spawn an isolated git-worktree worker for a rubric subset; integrate finished workers back.