repo-auditor

Pass

Audited by Gen Agent Trust Hub on Jun 20, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its workflow involves reading and processing untrusted data from git diffs and changed file contents. An attacker could embed malicious instructions within code comments or documentation in a pull request to influence the auditing agent's output or behavior.\n
  • Ingestion points: SKILL.md workflow step 1 (git diff) and file content reads.\n
  • Boundary markers: None provided; the agent is instructed to walk the diff and file content as part of its checklist processing.\n
  • Capability inventory: Local file reading, repository metadata access via git, and report generation.\n
  • Sanitization: The instructions do not specify any validation or escaping mechanisms for the ingested diff data.\n- [COMMAND_EXECUTION]: The skill's workflow suggests using the user-provided base input directly in a shell command (git diff <base>...HEAD). This creates a potential command injection surface if the input is not sanitized by the agent's tool execution environment before being passed to the shell.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 20, 2026, 03:37 PM
Security Audit — agent-trust-hub — repo-auditor