Harness Score — AI skill for Claude Code
Archon workflow that scores a repo's AI-agent readiness across 10 MVP checks (0-100).
How to install Harness Score
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open seanrobertwright/harness-score and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Harness Score does
Archon workflow that scores a repo's AI-agent readiness across 10 MVP checks (0-100). Deterministic detection + Claude Opus 4.7 synthesis.
Alternatives in AI
- Part 4 — Generate AGENTS.md And AI Agent Configuration Files — I'll help you create the instruction files that will guide your AI coding assistant to build your MVP 2.1k ★
- LongHorizon Harness — The long-horizon computer-use harness 1.3k ★
- Gtm Engineer Skills — Claude Code skill for improving website AEO (AI Engine Optimization) and GEO (Generative Engine Optimization) 761 ★
README
harness-score
An [Archon](https://github.com/coleam00/Archon) workflow that audits a repository and answers:
**How ready is this repo for reliable AI-agent-driven development?**
Produces a Harness Score out of 100 with a rating, per-check breakdown, and ranked fixes.
Based on the [Archon Harness Score Diagnostic System](https://github.com/coleam00/Archon/discussions) MVP spec.
What it does
Two nodes, ~25 seconds end-to-end:
detect(deterministic Bun script) — scans the repo for 10 MVP signals and emits scored JSON to stdout.report(Claude Opus 4.7) — reads the JSON and writes$ARTIFACTS_DIR/harness-score.mdwith rating, score table, and the top 3 fixes ranked by potential score gain.
Checks
| # | Check | Max | What it looks for |
|---|---|---|---|
| 1 | agent_instructions |
10 | AGENTS.md, CLAUDE.md, .cursorrules, or AGENT.md at repo root |
| 2 | readme |
10 | README.md, README.rst, README.txt, or README |
| 3 | build_test_commands |
10 | package.json / pyproject.toml / Cargo.toml / go.mod with build + test scripts |
| 4 | ci_workflows |
10 | .github/workflows/, GitLab CI, CircleCI, Azure Pipelines, Travis, or Jenkinsfile |
| 5 | test_directories |
10 | test/ / tests/ / __tests__/ / spec/ dirs or *.test.* / *.spec.* files |
| 6 | lint_typecheck |
10 | npm scripts, ESLint/Biome/Ruff/mypy/pyright config, or tsconfig.json |
| 7 | archon_workflows |
10 | .archon/workflows/ with at least one YAML file |
| 8 | security_scanning |
10 | Dependabot, Renovate, security CI workflow, or audit script |
| 9 | sensitive_paths |
10 | auth/ / billing/ / infra/ / secrets/ dirs (7pt) + .env in .gitignore (3pt) |
| 10 | markdown_report |
10 | Always passes when this workflow runs |
Rating bands
| Score | Rating |
|---|---|
| 90–100 | Agent-ready |
| 75–89 | Mostly ready |
| 60–74 | Partially ready |
| 40–59 | Risky |
| 0–39 | Not harnessed |
Install
Related Skills
Yotta Mirror
Deterministic, local learning-diagnostics skill for AI agents: score sheets plus an item-to-knowledge map beco
Council Refine
/council-refine — Multi-Model Plan Refinement with Opus 4.6 Synthesis
House Party Protocol
Evidence-first harness for coding agents across Claude Code and Codex CLI: cross-model review, governed loops,
DASA
Deterministic RAG: Agent A retrieval + Agent B grounded synthesis. CPU-only, SHARD-backed IVF-PQ retrieval tha
Select Coder
Recommend which coder agent and model should execute a coding task — probes locally available backends (opus,
Researching Web Skill
AI-powered web research skill for Claude Code. Generates HTML reports with confidence scores, contradiction de
Related Agents
Copyeditor
Sole owner of deterministic quantitative checks across all Layers. Runs the 10 tools/lint.py groups (graph·hub
Audit Content
Evaluates E-E-A-T signals, readability, thin content detection, duplicate content, freshness, and AI citation
Creo SEO
Next.js SEO + GEO specialist. Full-site audits across 7 phases covering technical SEO, content quality with AI