For people who build on Claude Code

Every skill hits a crossroads before you ship it.

Skill Crossroads grades your Claude Code skills, agents, and MCP servers against an evidence-based rubric — then points you one of three ways: ship, fix, or rethink.

Point it at a repo, a local folder, or your CI: npx skillcrossroads ./my-skill

A grade is a direction, not a gold star.

ShipA / B

Your artifact is solid. Embed the badge and release with confidence.

FixC / D

Close, but Skill Crossroads found specific problems — each with the file and line to change.

RethinkE / F

Deeper issues: it will not trigger, is not safe, or has no way to prove it works.

How it works

  1. 1

    Point Skill Crossroads at your artifact.

    npx skillcrossroads ./my-skill, a repo URL, or your CI.

  2. 2

    It runs the rubric.

    Fast deterministic checks plus AI-assisted review across six categories — every finding cited to a file and line.

  3. 3

    You get a scorecard, a badge, and a fix list.

    Ranked by how much each fix raises your grade.

What Skill Crossroads checks

Correctness & Structure

Valid frontmatter and manifest, resolvable references, nothing pointing at files that do not exist.

Triggering & Discoverability

Will the model actually invoke it? The number-one reason good skills look broken.

Clarity & Instructions

Unambiguous, contradiction-free, and phrased as standing instructions.

Token & Context Cost

What it costs every turn, and whether it uses progressive disclosure.

Safety & Security

Over-broad tool grants, injection surface, and secrets that should not be there.

Verifiability & Maintainability

Are there real evals, or tests that only grep the source?

Receipts, not vibes.

Every finding cites the file and line, and shows what your artifact claims versus what Skill Crossroads could verify. No hype, no false confidence — Skill Crossroads will tell you when your own skill scores a C, and exactly why.

Put the signpost in your README.

One line embeds your Skill Crossroads badge. Anyone who sees it can click through to the full, evidence-cited scorecard — and run their own. Good work gets shown; the badge does the rest.

[![Skill Crossroads: A−](https://skillcrossroads.com/api/badge/OWNER/REPO.svg)](https://skillcrossroads.com/s/OWNER/REPO)

Pricing

Free

Scan public artifacts and local files, full rubric, local badge. This is the whole tool, in the open.

Pro

Private repos, hosted always-fresh badges, score history, managed AI checks.

Team

CI quality gate on pull requests, org rules, shared dashboard.

Free covers everything you need to grade and share a public skill.

The State of Claude Code Skills.

Skill Crossroads publishes evidence-based reports on what actually makes real skills pass or fail across the ecosystem.

Read the report