suede-full-send
Pass
Audited by Gen Agent Trust Hub on Aug 18, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill uses brand-specific terminology like 'Full Send' to encourage high-effort reasoning and execution, but it explicitly mandates compliance with all host safety rules, spend limits, and privacy boundaries.
- [SAFE]: The 'Freeze The Mission' section requires the agent to establish a formal record of targets and authorized actions before any mutation occurs, creating a clear security and operational boundary.
- [SAFE]: A rigorous proof standard and reconciliation loop are implemented, requiring direct evidence (such as diffs, test results, and live environment checks) to verify all agent-reported outcomes, preventing unverified or deceptive completion claims.
- [SAFE]: The skill manages potential indirect prompt injection surfaces (user-supplied objectives and target surfaces) through structured mission documentation and explicit data handling rules.
- Ingestion points: User-defined objectives and targets entered into the mission record.
- Boundary markers: The structured mission schema and the 'Freeze The Mission' protocol.
- Capability inventory: Orchestration of specialized tools for project-wide mutation.
- Sanitization: Specific instructions for secret redaction and 'sensitive_source_rules' to protect private data.
Audit Metadata