The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches repository statistics and file contents from well-known GitHub domains (api.github.com and raw.githubusercontent.com) for the purpose of security auditing.
[COMMAND_EXECUTION]: Provides example shell commands for the agent to use curl and jq to retrieve and inspect metadata from external repositories.
[PROMPT_INJECTION]: Susceptible to indirect prompt injection (Category 8) as it is designed to ingest and process instructions from external sources during the vetting process.
Ingestion points: Shell commands in SKILL.md used to fetch remote skill files for inspection.
Boundary markers: None specified in the provided example commands; content is displayed directly for review.
Capability inventory: Access to curl and jq utilities for network retrieval and data parsing.
Sanitization: Relies on the agent following the vetting protocol to detect and handle malicious instructions within the fetched data.

skill-vetter