fetch-milestone-prs

Pass

Audited by Gen Agent Trust Hub on Jun 21, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Indirect Prompt Injection Surface: The skill retrieves titles, bodies, and labels from GitHub pull requests and saves them to the local file system.
  • Security Consideration: If the collected metadata is later used as input for other automated processes or LLM prompts (such as a release notes generator), there is a potential risk that malicious instructions hidden within a pull request body could influence the agent's behavior.
  • Evidence Chain: Untrusted data is ingested from the GitHub API in fetch-milestone-prs.py and saved to individual JSON files without sanitization or boundary markers.
  • Command Execution via Subprocess: The script executes the GitHub CLI (gh) to interact with the repository API.
  • Security Consideration: The use of subprocess.run allows the script to interact with the underlying system, which requires trust in the tool's logic and the environment where it executes.
  • Context: The commands are executed using argument lists rather than shell strings, which effectively mitigates common shell injection vulnerabilities.
  • File System Interaction: The skill creates directories and writes multiple JSON files to the disk based on user-provided milestone names and output paths.
  • Security Consideration: Automated file creation based on external input requires ensuring that the generated paths remain within the intended workspace boundaries.
  • Context: The implementation uses standard library path management to handle directory creation and file writing for PR metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 21, 2026, 10:25 AM
Security Audit — agent-trust-hub — fetch-milestone-prs