comparison
inference shell skills vs skills.sh
skills.sh uses git SHAs. inference shell skills are versioned, security-scanned, and evolving.
| skills.sh | inference shell | |
|---|---|---|
| skill discovery and browsing | yes | yes |
| cross-runtime (claude code, cursor, cline) | yes | yes |
| semantic versioning | - | yes |
| security scanning on publish | - | yes |
| verified namespaces | - | yes |
| lineage and evolution tracking | - | yes |
| fork and publish back | - | yes |
| windows support | - | yes |
| creator reputation system | - | yes |
the key difference
skills.sh proved the concept: markdown files that teach agents new capabilities, installable from a central directory. it works and it was first.
inference shell skills builds on that foundation with proper versioning (semantic versions, not just git SHAs), security scanning on every publish, verified namespaces (no squatting), and evolution tracking. when someone forks and improves a skill, the lineage is tracked and the best forks surface automatically.
security by default
every skill published to inference shell is scanned for prompt injection, credential leaks, exfiltration patterns, and unsafe tool calls. six automated checks run on every version before publish. skills.sh has no scanning. you trust the author entirely.
verified namespaces mean @anthropic is actually Anthropic. no squatting, no impersonation. creator reputation builds over time based on install counts and community trust signals.
yes. skills are markdown files. copy your skill content, upload with belt skill upload, and you get versioning, scanning, and a verified namespace on top.
ready to ship?
start with the hosted platform. deploy your own when you're ready.
we use cookies
we use cookies to ensure you get the best experience on our website. for more information on how we use cookies, please see our cookie policy.
by clicking "accept", you agree to our use of cookies.
learn more.