muse-delegate

Pass

Audited by Gen Agent Trust Hub on Oct 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a Node.js relay script to execute local muse and git commands. These tools are used for the primary purpose of the skill: performing coding implementation and monitoring repository state changes during the delegation process.
  • [DYNAMIC_EXECUTION]: The relay.mjs script contains logic to locate and execute a sibling script, lane.mjs, when specific configuration lanes are used. This execution is confined to local files within the skill's deployment directory and follows standard module-level interaction patterns.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed for an orchestrator agent to process output generated by a secondary implementer agent. This creates a surface for indirect instructions to influence the orchestrator. The skill proactively mitigates this risk by providing detailed review checklists and instructing the orchestrator to manually re-run project validation gates rather than trusting self-reported outcomes.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 13, 2026, 08:08 PM