find-skills
Pass
Audited by Gen Agent Trust Hub on Feb 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches skill metadata from the
skills.shregistry and downloads package content from GitHub repositories. It references both trusted organizations likevercel-labsand third-party sources. - [COMMAND_EXECUTION]: Utilizes the
npx skillsCLI for package management, including the use of the-yflag to bypass confirmation prompts. - [REMOTE_CODE_EXECUTION]: Facilitates the download and execution of external scripts from remote repositories as its primary purpose.
- [PROMPT_INJECTION]: The processing of untrusted search results from an external registry creates a surface for indirect prompt injection. 1. Ingestion points: Output from the
npx skills findcommand. 2. Boundary markers: None; search results are presented directly. 3. Capability inventory: Shell execution of thenpx skillstool suite. 4. Sanitization: No validation or filtering is performed on external metadata before presentation or use.
Audit Metadata