chatcrystal-task-writeback
Pass
Audited by Gen Agent Trust Hub on Apr 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or safety bypasses were detected in the skill files.
- [DATA_EXFILTRATION]: The skill transmits project context including directory paths, branch names, and work summaries to a persistence tool. This is the intended and documented behavior for recording task memories within the ChatCrystal ecosystem.
- [REMOTE_CODE_EXECUTION]: The skill explicitly includes instructions to prevent the agent from automatically installing external software, which mitigates the risk of unauthorized code execution during setup.
- [PROMPT_INJECTION]: The skill uses natural instructional language and does not contain patterns associated with overriding agent safety guidelines or system instructions.
Audit Metadata