pr-risk-scoring

Pass

Audited by Gen Agent Trust Hub on Mar 21, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses gh pr list and gh pr diff to retrieve pull request data for analysis. These are standard GitHub CLI operations.
  • [COMMAND_EXECUTION]: A Python script score_prs.py is executed locally to process the JSON and CSV data retrieved from GitHub. This script performs heuristic-based scoring and does not make external network requests.
  • [DATA_EXPOSURE]: The skill generates a CSV report containing pull request metadata (titles, authors, and risk scores) at todo/pr_risk_scores.csv. This data is derived from the repository's own public or internal pull request history.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data in the form of PR titles and author names which are then included in a generated CSV and displayed in the agent's output. While this presents an injection surface, the risk is minimal as the data is used for display and scoring purposes only, and the scoring logic uses strict regex patterns for filenames rather than interpreting natural language content.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 21, 2026, 06:35 AM
Security Audit — agent-trust-hub — pr-risk-scoring