repo-explore

Pass

Audited by Gen Agent Trust Hub on Mar 26, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted data from external Git repositories.
  • Ingestion points: Untrusted content enters the agent's context through the git clone command and is subsequently read using tools like rg, find, and ls (SKILL.md).
  • Boundary markers: The instructions do not define clear boundaries or provide warnings to the agent to ignore potential instructions found within the codebase being explored.
  • Capability inventory: The agent has access to several shell tools (git, rm, find, rg, etc.) which could be targets for manipulation if the agent's logic is subverted by malicious repository content.
  • Sanitization: There is no evidence of sanitization or filtering applied to the content of the cloned files before they are presented to the agent.
  • [EXTERNAL_DOWNLOADS]: The skill is designed to download external source code from user-provided URLs using the git clone command. This is a core feature of the skill but involves fetching content from untrusted remote sources.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 26, 2026, 09:14 AM
Security Audit — agent-trust-hub — repo-explore