Skip to content
SkillScript-led1,093 tokens in use

Dual Axis Skill Reviewer

Scores agent skills on two axes, deterministic code checks and an LLM review, then merges them into a weighted score with improvement items.

7,426 stars on the repository MIT Licensetradermonty/claude-trading-skills · skills/dual-axis-skill-reviewer

Skill label

measured from its files

What installing Dual Axis Skill Reviewer costs

96tokens idle
1,093tokens in use
1,703tokens, up to

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Up to adds the files its instructions link. Counts are approximate, made offline with cl100k_base.

Model dependence Script-led

Script-ledMixedModel-led

Most of this skill is scripts, so it behaves much the same whichever model runs it.

This describes the skill's structure, how much of it is scripts. It does not say how well it works. Scripts: scripts/run_dual_axis_review.py.

Touches

shelldetectedwrites filesdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Good
  • Frontmatter12/15
  • Structure14/25
  • Examples12/15
  • Output10/15
  • Size15/15
  • Resources7/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent
  • Runs shell commands−5

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 43
  • Writes files−8

    does this on the machine, and the frontmatter does not say so

    scripts/run_dual_axis_review.py, line 1061

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity100
  • Maintenance52
  • Community68
  • Maturity36
  • Momentum40

Dual Axis Skill Reviewer scores skills in a project’s skills folder. The first axis is computed by a Python script from the skill’s files: metadata and use case, workflow coverage, execution safety, supporting artifacts and test health. The second axis is a review written by an LLM from a prompt the script generates, covering correctness, risk, missing logic and maintainability.

The script merges the two with adjustable weights into a final score, and lists improvement items when the score falls below 90. A project root option lets it review skills in other projects. This repository’s skill tooling is the subject here, not market analysis.

What You Get

  • A JSON report and a Markdown report per skill with auto score, LLM score and final score
  • An LLM review prompt file generated for each skill
  • Improvement items listed whenever the final score is under 90
  • A display of the declared verification block from skills-index.yaml, shown for information and left out of the scores

Common Use Cases

  • Gating merges of new skills on a minimum score such as 90
  • Getting concrete fixes for a low-scoring skill
  • Reviewing skills in a different project with the project root option
  • Quick triage of every skill in a project with tests skipped

What Makes It Unique

Reproducible scoring and qualitative review are kept as separate axes with explicit weights, so a team can lean toward the deterministic gate or toward the deeper review. Skills with no scripts are detected as knowledge-only and are not penalized for lacking tests. Seeds and named skills make runs repeatable.

Under The Hood

The agent runs one Python script with uv, which resolves the pyyaml dependency. The script reads the target project’s skills folder and files, and unless told to skip tests it can run the reviewed skill’s test suite, which executes code from that project. It writes JSON and Markdown reports and a prompt file into an output folder (reports/ by default). It makes no network calls of its own. The skill also suggests a shell command that symlinks the skill folder into the user’s Claude skills directory for global use.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers