implement-issue

Pass

Audited by Gen Agent Trust Hub on Jul 24, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes external data from GitHub issues, comments, and related PRDs, creating a surface for indirect prompt injection.
  • Ingestion points: External content is fetched using gh issue view and read into the agent's context (SKILL.md).
  • Boundary markers: The skill does not explicitly define markers to separate untrusted issue content from instructions, though it relies on standard issue-viewing behavior.
  • Capability inventory: The agent can modify local files and execute commands such as git, gh, and uv (SKILL.md).
  • Sanitization: No explicit sanitization or filtering of the fetched issue content is performed, placing reliance on the underlying model's guardrails.
  • [COMMAND_EXECUTION]: The skill relies on established local command-line tools to interact with the repository and validate code changes.
  • Evidence: Commands include git (status, log, diff, commit), gh (issue view), and uv (python -m compileall) (SKILL.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 24, 2026, 09:04 AM
Security Audit — agent-trust-hub — implement-issue