find-skills

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses the npx skills command to interact with the open agent skills ecosystem registry. It facilitates the download and installation of modular skills from external sources, specifically referencing official repositories from well-known organizations like Vercel Labs.
  • [COMMAND_EXECUTION]: The provided scripts/install-skill.sh script executes shell commands to automate the installation process. It utilizes npx skills add to install packages and standard filesystem operations (mkdir, ln) to link installed skills into the local project environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill functions as a discovery engine that ingests search results from an external registry. This creates an attack surface where an agent might be influenced by metadata or instructions found within third-party skills returned in search results.
  • Ingestion points: Search results and skill descriptions returned by npx skills find in SKILL.md.
  • Boundary markers: The skill provides structured response templates to clearly separate skill metadata from installation instructions.
  • Capability inventory: The skill includes a shell script (install-skill.sh) capable of executing npx and managing symlinks.
  • Sanitization: The installation script implements validation to ensure skill identifiers match the expected owner/repo@skill-name format before proceeding with installation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 02:01 PM