standup
Pass
Audited by Gen Agent Trust Hub on Jun 12, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a bundled Node.js utility ('standup.mjs') to execute standard git and GitHub CLI ('gh') commands for branch and PR discovery. These operations are restricted to gathering metadata necessary for the standup task and do not involve arbitrary command execution.
- [COMMAND_EXECUTION]: The facilitator agent spawns sub-agents using the 'Task' tool, passing them a specific instruction set ('agent-brief.md') to participate in the standup rounds.
- [PROMPT_INJECTION]: The skill uses instructional role-play to define a 'facilitator' agent that coordinates 'sub-agents'. This is a structural design for agent-to-agent collaboration and does not attempt to bypass security filters or override system constraints.
- [DATA_EXFILTRATION]: No sensitive data is transmitted externally. The skill reads local git metadata and public PR information to provide context for the standup conversation.
- [SAFE]: The architecture relies on a local markdown-based 'chat room' with an atomic locking mechanism to prevent race conditions. The 'read-only' instruction for agents during the reconciliation phase is a security best practice that defers actual file modifications to the user's explicit oversight.
Audit Metadata