refresh-skills
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches content from the author's official GitHub repositories (philtzjp/skills and philtzjp/startingpoint) to update guidelines and compare local skill modifications. These downloads are restricted to the vendor's own infrastructure.
- [COMMAND_EXECUTION]: The skill uses shell commands such as npx skills, curl, ln, and diff to manage the installation and configuration of skills. The instructions specify using DISABLE_TELEMETRY=1 and require user confirmation before destructive actions like deleting files or applying updates.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external SKILL.md files during comparison and migration tasks.
- Ingestion points: Remote skill definitions are retrieved via curl, and local files in .agents/skills/ are read for comparison.
- Boundary markers: The skill relies on manual verification of diff outputs rather than automated delimiters within the files.
- Capability inventory: Shell execution (npx, curl, ln, diff) and file system manipulation.
- Sanitization: The workflow enforces human-in-the-loop validation, explicitly instructing the agent to show differences to the user and obtain permission before proceeding with deletions or migrations.
Audit Metadata