The Agent Skills Directory

[SAFE]: The skill defines a structured workflow for task verification using independent subagents. It includes robust security practices such as filesystem isolation for reviewers, SHA-256 hashing to ensure candidate integrity, and explicit prohibitions against sharing context or findings between reviewers. No malicious patterns, data exfiltration, or unauthorized command executions were detected. The skill uses standard agent capabilities to manage subtasks and manage a local 'ledger' of results within the project workspace.

adversarial-review