CHAINGUARD AGENT SKILLS
Don’t let your AI agent talk to strangers
Agents are only as safe as the skills you give it. Stop letting it pick up unvetted dependencies from the digital sidewalk. Use Chainguard’s catalog of hardened, safe-to-use skills instead.
Directly address agent skill attack vectors with rulesets targeting real-world attack patterns, such as unrestricted shell access and overly broad permissions.
Developers install a hardened skill by simply copying a single SKILL.md file. No new toolchain or CI integration required.
When upstream sources change or rules are updated, every skill is automatically re-evaluated and re-hardened.
Chainguard Agent Skills
Sign up for the beta to use Agent Skills
Hardened public skills
Pull from Chainguard's public registry instead of the open marketplace. Every skill is scanned and fixed with a full hardening report so you know exactly what was checked and what was remediated.
Your skills, hardened by the Chainguard Factory
Bring your own skills. Chainguard hardens them in the same Factory that secures every Chainguard product.
Private registry with access control
Give your internal skills a real home. Manage who can push and pull, pin agents to specific skill versions, and keep everything inside your org boundary.