default

Pass

Audited by Gen Agent Trust Hub on Jul 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a 'design-first' approach in the Brainstorming phase, requiring the agent to obtain explicit user approval for a proposed plan before proceeding to execution.
  • [COMMAND_EXECUTION]: The workflow involves file system operations, specifically creating task summaries in a './docs/' directory and deleting temporary tracking files ('task_plan.md', 'findings.md', 'progress.md'). The instructions include explicit checks to prevent the deletion of pre-existing user files.
  • [SAFE]: The skill delegates sensitive operations like GitHub repository access to a specialized 'skill-standard-harness', which centralizes the management of access protocols and security standards.
  • [SAFE]: Assessment of Indirect Prompt Injection surface: 1) Ingestion points: Untrusted user task descriptions are processed at the start of the workflow in SKILL.md. 2) Boundary markers: The mandatory Brainstorming phase acts as a manual checkpoint. 3) Capability inventory: File system read, write, and deletion operations are utilized during the Planning and Archiving phases. 4) Sanitization: The skill relies on human-in-the-loop review and approval of the generated plan before any substantive actions are taken.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 18, 2026, 07:41 AM
Security Audit — agent-trust-hub — default