intaking-requirement
Pass
Audited by Gen Agent Trust Hub on Jun 24, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill employs a robust governance framework through an action classification system (A/R/N) and mandatory auditing, which ensures that sensitive actions like card creation and status changes follow established authorization protocols.
- [COMMAND_EXECUTION]: The routine utilizes a local helper script (
scripts/lib/common.sh) to generate creator metadata traces. This operation is restricted to local utility functions for internal auditing and does not involve the execution of untrusted external content. - [SAFE]: Indirect prompt injection risks are effectively mitigated through the skill's mandatory intake pipeline. Ingestion points: User-provided requirements, ideas, and bug reports described in the triggers of
SKILL.md. Boundary markers: A required acknowledgment step (Step 1) where the agent must confirm the scope and intent with the user before processing data. Capability inventory: The skill can create project cards and route requests to authorized sibling plugins (gstack:*,superpowers:*). Sanitization: The 'Spec-first checklist' and 'Shape judgment' gatekeepers ensure that all inputs are validated against the project's architectural premises and structural requirements before any state changes occur.
Audit Metadata