suggest-pr-reviewers

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The scripts/reviewers and tests/backtest scripts execute external binaries git and gh (GitHub CLI) via subprocess.run to interact with the repository history, configuration, and the GitHub GraphQL API.
  • [DYNAMIC_EXECUTION]: The tests/backtest script uses importlib.machinery.SourceFileLoader to dynamically load and execute the reviewers script from a file path that can be provided as a command-line argument (--script).
  • [DYNAMIC_EXECUTION]: The tests/test_reviewers.py script generates a mock GitHub CLI tool at runtime by writing a Python source string (FAKE_GH) to a temporary file and executing it via subprocess during unit testing.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the repository being analyzed, such as Git commit subjects and GitHub pull request metadata (titles, reviewer logins). This information is processed and displayed to the AI agent in a ranked table, creating a surface where a malicious contributor could attempt to influence the agent's behavior through crafted repository content.
  • Ingestion points: Git log output (commit subjects, author names) and GitHub API responses (PR metadata).
  • Boundary markers: The output is structured in a Markdown table, but it lacks explicit sanitization or escaping for the potentially adversarial content within individual cells.
  • Capability inventory: The skill has the ability to read local files and execute shell-based tools like git and gh.
  • Sanitization: While the script filters for known bot accounts, it does not sanitize user-controlled strings like commit subjects before presenting them to the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 01:37 PM