Audit
Harness engineering tools for AI agents — audit, init, recommend your project's agent-readiness (AGENTS.md, context engineering, eval)
npx -y skills add rulebased-io/harness --skill auditAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Audits your project's harness engineering setup and assigns a score (36 items across 8 categories, 3-level grading)
SKILL.md
1.7 KB, as published. Nobody here has run it
Audits the current project's harness engineering setup.
The CLAUDE_PLUGIN_PATH provided by the hook is this plugin's root. Read ${CLAUDE_PLUGIN_PATH}/reference/index.md for the full 36-item criteria, scoring, and hollow detection rules.
Reading order: index.md → guide-*.md (methodology) → *.md (item details, only if needed)
How to Audit
- Read
${CLAUDE_PLUGIN_PATH}/reference/index.mdfor the scoring system, 36-item summary, and hollow criteria - For each category, read the
guide-*.mdfile for evaluation methodology - Check each item against the current project:
- pass: Structure exists AND has meaningful content
- hollow: Structure exists but empty/placeholder only (see
Hcolumn in index.md) - fail: Structure does not exist
- Calculate the weighted score using the 3-level formula
Scoring
- Critical (weight 3) · Important (weight 2) · Nice-to-have (weight 1)
- Formula:
score = (pass_weights + hollow_weights × 0.5) / total_weights × 100 - Grades: A (90+), B (75+), C (60+), D (40+), F (<40)
Output Format
[PASS] ctx-agents-exists — AGENTS.md exists (3/3)
[HOLLOW] ctx-agents-build — Build commands section exists but no actual commands (1.5/3)
→ AGENTS.md의 빌드 섹션에 실제 npm/yarn/pnpm 커맨드를 추가하세요
[FAIL] ctx-agents-arch — No architecture description (0/2)
→ AGENTS.md에 프로젝트 구조 트리와 디렉토리 설명을 추가하세요
End with: overall score, grade, and top 3 improvement recommendations.
$ARGUMENTS