investigate-codebase

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to locate and follow repository-specific instruction files like CLAUDE.md or AGENTS.md. This behavior creates an attack surface where a malicious codebase could provide conflicting or harmful instructions to the agent during an investigation.
  • Ingestion points: File reading operations target instruction and configuration files within potentially untrusted code repositories as described in SKILL.md.
  • Boundary markers: The skill lacks mechanisms to distinguish between safe system-level instructions and untrusted instructions found in data files.
  • Capability inventory: The agent uses search (rg, grep) and execution tools which could be misused if hijacked by malicious repository instructions.
  • Sanitization: No validation or sanitization of embedded instruction content is described in the provided references.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 05:34 AM
Security Audit — agent-trust-hub — investigate-codebase