Free
$0
Browse skillsWe only list skills we ran.
Most skills on GitHub do not work. Each review here is an A/B test: the same tasks, runs without the skill and with it, scored by rule-based checks. A Codex pass covers Claude Code.
Tier
1 SOTA agents / Codex
Claude Code
2 Mid-tier agents / DeepSeek Flash (re-testing)
Coming later
Top picks
All skills29.0
Total /30 · Lift +3.33
pr (mattpocock/skills)
pr: a pure format skill: control never produces the three-section PR body (1.7/5), the skill and a one-sentence prompt both do (5.0 and 4.95), at no extra tokens
RecommendedSafe
SOTA Pass
29.0
Total /30 · Lift +3.59
setup-pre-commit (mattpocock/skills)
setup-pre-commit: for a vague 'add commit hooks' request plain Codex builds a different (native git hook) setup, even calling npm in a pnpm repo (seen in one run); the skill and a one-sentence prompt both give the Husky + lint-staged setup (5.0 vs 1.4)
RecommendedSafe
SOTA Pass
28.2
Total /30 · Lift +2.20
domain-modeling (mattpocock/skills)
domain-modeling: turns plain 'set up a glossary / record this decision' requests into the short GLOSSARY.md + ADR format (control 2.8, skill 5.0), a one-sentence prompt gets 4.6; no token overhead
RecommendedSafe
SOTA Pass
Browse by category
Two tiers, one protocol
Every skill is tested on a frontier agent first. A pass on the test agent counts for the others in that tier. A second, cheaper-model tier (DeepSeek Flash) is being re-tested and is marked untested until then.
| Tier | Tested with | A pass also covers |
|---|---|---|
| SOTA agents | Codex | Claude Code |
| Mid-tier agents | DeepSeek Flash (re-testing) | Coming later |
Free and Pro
Read every published review for free. Pro unlocks the raw test record.
Pro
$79/ year
$6.58 / month, billed yearly
| Included | Free | Pro |
|---|---|---|
| Verdict and scores | Yes | Yes |
| Written review | Yes | Yes |
| Sample artifact | One text sample | Every artifact |
| Sandbox log | First 12 lines | Full log |
| Every Codex output file | Not included | All of them |
| Original video | Not included | Included |
| Security report | Conclusion only | Full report |
| Per-run scores | Not included | Every run |
| Re-test alerts | Not included | Included |
Questions
What can I read without Pro?
The verdict, scores, written review, and security conclusion are free. You also see the first 12 lines of the sandbox log and one sample artifact. Pro adds the full log, every Codex artifact, the original video, and the full security report.
How do I cancel?
Open Billing and choose Manage billing. Cancellation happens in the Stripe customer portal. You keep Pro until the end of the period you already paid for.
What happens if a charge is refunded?
When a refund is recorded, Pro ends. The free review stays on the site. A subscription you cancel without a refund keeps access until the period ends.
What does the subscription license?
Pro lets you read this site's test record. It is not a license to the skill itself, and it does not allow you to republish the logs, video, or artifacts.
Questions
Have a skill worth running?
Send the repository. We test it on both tiers before it can appear here.