Security scan
Run SkillGuard before installing third-party Agent Skills, with optional semantic review only when the user explicitly allows network/model use.From its SKILL.md
npx -y skills add stella/skillguard --skill security-scanAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
3 things to look at
- reads credentialsReads from 1 credential source: `OPENAI_API_KEY`.
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
- runs commandsInstructs the agent to run 3 commands, including `skillguard scan <skill-path> --preset strict --policy strict` and 2 more.
SKILL.md
1.2 KB, 233 tokens by cl100k_base, as published. Nobody here has run it
SkillGuard Security Scan
Use this skill before installing or enabling a third-party Agent Skill.
Default Static Scan
Run a deterministic local scan first:
skillguard scan <skill-path> --preset strict --policy strict
Do not install the skill if the scan fails unless the user explicitly accepts the risk after reviewing the findings.
Semantic Review
Run semantic review only when the user explicitly approves sending the candidate skill content to a model provider:
skillguard scan <skill-path> \
--preset paranoid \
--policy strict \
--allow-network \
--semantic \
--semantic-provider openai
Semantic review requires provider credentials, normally OPENAI_API_KEY. For
OpenAI-compatible local endpoints, use:
skillguard scan <skill-path> \
--preset paranoid \
--policy strict \
--allow-network \
--semantic \
--semantic-provider openai-compatible \
--semantic-base-url http://localhost:11434/v1
Never include unrelated user documents, chat history, workspace secrets, or credentials in the scan target.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.