review

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION] (HIGH): Vulnerable to Indirect Prompt Injection (Category 8). The skill ingests untrusted external data which can influence its behavior.
  • Ingestion points: Fetches data from external sources using gh pr view and gh issue view (Phase 1, Step 2).
  • Boundary markers: None. The skill does not use delimiters or instructions to ignore embedded commands within the fetched PR/Issue content.
  • Capability inventory: Uses gh CLI for network/API interaction and runs "available linting tools" (Phase 2, Step 3), which involves executing shell commands/binaries.
  • Sanitization: None. The skill directly processes and "reviews" the untrusted content.
  • [COMMAND_EXECUTION] (HIGH): The skill invokes shell commands (gh) and triggers the execution of linting tools based on "detected languages." If a linter is configured via a file in the repository (e.g., a malicious config file), it could lead to arbitrary code execution when the agent attempts to run it.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 09:01 AM