conversation-review
Pass
Audited by Gen Agent Trust Hub on May 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's primary purpose is meta-analysis of conversation history, which is a benign functional capability. The 'Dolores' mode serves as a structured framework for self-reflection.
- [SAFE]: Explicit privacy protections are documented within the core process and safety sections, specifically instructing the agent to 'Do not preserve raw private conversation text' and 'Abstract sensitive examples before proposing evals'.
- [SAFE]: The skill defines local file reading paths (e.g.,
cases/**/*.md) for generating quality dashboards, which is a standard diagnostic behavior for maintaining a feedback loop within a developer environment and does not involve exfiltration. - [SAFE]: No network operations, external downloads, obfuscation, or arbitrary command execution patterns were found.
Audit Metadata