paladini/harness-score

What it solves

AI coding agents (like Cursor, Claude Code, or Windsurf) are often unreliable because they lack a structured "harness"—the set of rules, guides, and guardrails that prevent them from making mistakes or rediscovering project conventions every session. Harness Score provides a deterministic way to measure the maturity of this infrastructure, moving beyond "vibe-coding" to a quantifiable, engineering-based approach to agent reliability.

How it works

It is a deterministic scanner that analyzes a repository's filesystem without using LLMs or network access. It runs 36 checks across six dimensions (Context & Guides, Skills & Commands, Hooks & Guardrails, Sensors & Feedback, CI Feedback, and Hygiene & Safety) to assign a maturity level from L0 (Unharnessed) to L4 (Self-correcting). The tool identifies missing artifacts—such as AGENTS.md files, .cursor/rules/, or CI pipelines—and provides a ranked list of specific fixes to reach the next maturity level.

Who it’s for

Developers and teams using AI coding agents who want to ensure their project is agent-ready and reliable, as well as DevOps engineers who want to gate their CI pipelines based on a minimum agent-harness maturity level.

Highlights

• Deterministic Scoring: Zero LLM calls and zero network requests ensure consistent results across different environments.
• Maturity Ladder: A 5-level system (L0-L4) that gates progress based on the shape of the harness rather than just total points.
• CI Integration: Includes a GitHub Action and a --min-level flag to fail builds if the agent harness regresses.
• Tool Agnostic: Works across various AI coding tools including Cursor, Claude Code, Windsurf, Cline, and Continue.
• Actionable Diagnosis: Every failed check links to a specific remediation recipe to fix the gap.