skills/phrazzld/agent-skills/groom/Gen Agent Trust Hub

groom

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (gh) to list, create, and edit issues, milestones, and projects. It also employs standard shell utilities such as find, cat, head, and mkdir to analyze local project context. The open command is used to display a locally generated HTML visual dashboard.
  • [EXTERNAL_DOWNLOADS]: The skill orchestrates sub-agents to perform web-grounded research and cross-repository investigation. These tasks target well-known technology organizations and utilize official GitHub APIs for data retrieval.
  • [DATA_EXFILTRATION]: The skill reads project-specific metadata files like project.md, CLAUDE.md, and .groom/retro.md to establish context for the session. This data is processed locally to inform the grooming process and is not exfiltrated to unauthorized domains.
  • [PROMPT_INJECTION]: The skill ingests external data from existing GitHub issues. While this constitutes an indirect prompt injection surface, the risk is mitigated by the skill's design, which requires multi-model consensus (/thinktank) and automated quality gates (/issue lint) before finalizing any output.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 12:38 PM