Practical patterns, starters & CLI tools for loop engineering with AI coding agents. Design systems that prompt and orchestrate agents (inspired by Addy Osmani and Boris Cherny). Includes loop-audit, loop-init, loop-cost.
TypeScript agentic-aiai-agentsai-codinganthropicautomationclaudeclaude-codecodexcoding-agentsdevops-automationdevtoolsgithub-actionsgrokllmloop-engineeringmcpprompt-engineering
Changelog
All notable changes to @cobusgreyling/loop-audit are documented here.
[Unreleased]
Changed
- Placeholders no longer score. Empty or whitespace-only files, and
{}/[]JSON, earn nothing and are listed under Not counted. A repo of 19 empty files (142 bytes) used to score 100/L3; it now scores 72/L1 - A skill counts only with a
SKILL.mdcarryingname+descriptionfrontmatter (bare directories used to count); the same for Claude Code verifier agents gate.yamlmust declareversion: 1and adenylist:; oneloop-gatewould refuse is a failure, not a signal.github/and workflows count only when they contain a non-empty file- L3 requires proven guardrails: a
loop-drill.json(fromloop-drill --record) showing the currentgate.yamlpassing its drills in both directions, with no recorded guardrail failing. Repos that were L3 on files alone drop to L2 until they record
Added
signals.proof: guardrails that are proven, failed, untested or stale, read fromloop-drill.json- A guardrail that
loop-drillshows failing loses its points: gate →gateYaml, verifier →verifier(Verifier Theater), breaker → stall detection
[1.9.0] - 2026-08-28
Changed
- Loop activity requires a fresh
Last run(≤14 days), a datedloop-run-log.mdrow, or a git commit that actually touches state / run-log files LOOP.mdcadence text and atriagecommit on README no longer count as activity- Activity is worth 14 points (was 6)
- Missing
STATE.mdis a warning, not a fail, when.github/workflows/thin-loop.ymlis present - Companion funnels are not pushed to the top of
--suggest
[1.8.0] - 2026-07-29
Added
--auto-fixself-heal for missing repository structure (STATE, LOOP, budgets, gate.yaml, etc.)- Memory-engineering readiness signals and
--with-memoryrecommendations
[1.7.0] - 2026-07-20
Added
- Harness Runtime scoring signals for the LE → harness-foundry funnel:
.foundry/stack.yaml,stack.lock, sessions/traces, outerloop emit path, host integrate hints
- Loop Ready 80+ recommendation / human report CTA →
loop-init --with-foundry --suggestcopy commands for Foundry scaffold + firstfoundry run
[1.6.0] - 2026-07-09
Added
- Governance scoring signals: least-privilege tool scope, stall / no-progress detection, human-escalation path
allowed-toolsfrontmatter inSKILL.mddetected as a least-privilege signal--suggestrecommendations when governance signals are missing
[1.5.3] - 2026-07-06
Added
- Contributor quickstart CTA after CLI runs (skipped for
--json,--md, and--badge)
[1.5.2] - 2026-06-30
Added
- Detect verifier agents in
opencode.jsonandopencode.json.example(maker/checker split for Opencode loops) --suggestcopy commands for Opencode (loop-init --tool opencode)
[1.5.0] - 2026-06-30
Added
loop-constraints.mdandloop-constraintsskill detection in readiness scoring (+6 points when both present)- Recommendations when constraints file or skill is missing
[1.4.1] - 2026-06-13
Changed
- Updated package description and keywords for better discoverability on npm / npx (emphasizes "loop engineering", coding agents, and concrete usage examples).
[1.3.0] - 2026-06-09
Added
- Unit tests for scoring logic (
test/auditor.test.ts) --suggestnow mentionsloop-initscaffold CLI- Registry and starter coverage in audit recommendations
Changed
- CI gates on test suite before publish
[1.2.0] - 2026-06-09
Added
--suggestcopy-from-template commands for Grok, Claude Code, and Codex- Expanded signals: MCP, worktree evidence,
patterns/registry.yaml - L3 scoring threshold with verifier + state requirements
[1.1.0] - 2026-06-08
Added
--mdmarkdown report format- Safety doc and GitHub workflow detection
[1.0.0] - 2026-06-07
Added
- Initial Loop Readiness Score CLI (L0–L3)
--jsonoutput for CI integration