github-issues-manager
Pass
Audited by Gen Agent Trust Hub on Feb 26, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies and processes GitHub issues, which are external and untrusted inputs that could contain malicious instructions.
- Ingestion points:
scripts/monitor_issues.pyandscripts/generate_fix.pyfetch and read issue titles and bodies via the GitHub API. - Boundary markers: None detected. The response templates in
SKILL.md(e.g.,{{extracted_description}}) interpolate raw issue content directly into the agent's output context. - Capability inventory: The skill can perform GitHub actions such as posting comments, adding labels, and generating code/PR templates.
- Sanitization: No sanitization or escaping of issue content is implemented to prevent the agent from accidentally following instructions embedded in issue reports.
- [EXTERNAL_DOWNLOADS]: The scripts require several external Python dependencies (PyGithub, requests, matplotlib, pandas) to function. These are standard libraries for GitHub integration and data reporting but should be verified during installation.
Audit Metadata