task-clarifier
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a structured 'Execution Loop' that restricts the agent from searching or executing tasks until core user requirements (Outcome, Constraints, Acceptance) are resolved.
- [SAFE]: Features a 'Privacy Gate' policy that explicitly prohibits the inclusion of secrets, credentials, API keys, or private file paths in external search queries.
- [SAFE]: Defends against indirect prompt injection (Category 8) via an 'Untrusted Evidence Boundary' instruction that tells the agent to treat external content as data only and never as an override for user instructions.
- [SAFE]: Imposes a human-in-the-loop requirement for high-risk operations, such as publishing or deleting, requiring explicit confirmation even after task alignment is achieved.
- [SAFE]: Does not contain any executable scripts, remote downloads, or third-party package dependencies, relying solely on instructional content and existing agent capabilities.
Audit Metadata