bulk-github-skills-downloader

Warn

Audited by Gen Agent Trust Hub on Aug 29, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill is designed to discover and clone multiple GitHub repositories from unverified third-party sources (e.g., mrgoonie/claudekit-skills, obra/superpowers). This involves downloading external executable instructions and directory structures from sources outside the vendor's control.
  • [INDIRECT_PROMPT_INJECTION]:
  • Ingestion points: The skill ingests untrusted Markdown and script content by cloning remote GitHub repositories and extracting files into the agent's /skills/ directory.
  • Boundary markers: There are no documented boundary markers or 'ignore embedded instructions' warnings for the downloaded content to prevent the agent from obeying malicious instructions hidden in the third-party skills.
  • Capability inventory: The skill uses git clone to pull remote data and performs file system operations to 'flatten' and move downloaded content into active skill paths.
  • Sanitization: The skill employs filtering based on metadata (stars, age, license type) but does not perform security sanitization or code analysis of the content being ingested.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 29, 2026, 08:16 PM
Security Audit — agent-trust-hub — bulk-github-skills-downloader