domain-analysis
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill employs emphatic instructions like 'MANDATORY IMPORTANT MUST' to prioritize internal task management and user validation, which is consistent with its intended business logic and does not constitute a safety bypass.
- [PROMPT_INJECTION]: An indirect prompt injection surface is present as the skill ingests data from local files in 'plans/' and 'team-artifacts/'. Ingestion points: Project documentation read in Step 1. Boundary markers: Absent. Capability inventory: Task creation, file modification, and web research. Sanitization: Absent. This surface is assessed as safe due to the skill's restricted architectural scope and the requirement for human verification of all findings.
- [SAFE]: The skill does not attempt to access sensitive system paths, hardcode credentials, or execute code from untrusted remote sources.
Audit Metadata