SKILL.md:1the skill works → no evals/, tests/, or verification section found
No way to prove the skill does what it claims.
Fix Add an evals/ folder (test prompts + expected outcomes) and a '## Verify' step in the body.
overall
Correctness & Structure
100
pass
Triggering & Discoverability
85
1 warn
Clarity & Instructions
100
pass
Token & Context Cost
77
3 warn
Safety & Security
100
pass
Verifiability & Maintainability
53
1 fail
Top fixes — ranked by score impact
SKILL.md:1description will match user requests → no explicit cue ('use when …', 'triggers on …', quoted user phrases) found
Pure what-it-does prose — the model has to guess when to invoke it.
Drive an open GitHub PR to merge-ready state via a non-blocking state machine…
Fix State the trigger explicitly: 'Use when the user says "…", "…", or wants to …'.
SKILL.md500-line guidance → 661 lines
SKILL.md is 661 lines — over the ~500-line guidance. ~9,098 tokens (rough est.); ~$0.0273 per load at $3/Mtok reference rate.
Fix Move heavy reference material into supporting files and link to them (progressive disclosure), keeping SKILL.md lean.
SKILL.mdeverything inlined in SKILL.md → 646 body lines, 0 supporting files
Large SKILL.md (646 lines) with no supporting files — all of it loads every time the skill fires.
Fix Move heavy reference material into supporting files and link to them, so the model loads it only when needed.
SKILL.mda one-time authoring cost → ~9,098 tokens (rough est.) per invocation
Every invocation carries this: ~9,098 tokens (rough est.) per invocation — ~$27.29 per 1,000 invocations at the $3/Mtok reference rate.
Fix Cut the entry file below ~8k tokens — move reference material into supporting files so every future invocation gets cheaper.
★ scores higher than ≈20% of 208 public skills (deterministic rubric v1.2 sample, 2026-09)
Embed this badge
Always-fresh — it re-scans and updates on its own.
Grade your own Claude Code skill — evidence-cited, file-and-line, free.
Scan your own skill →