spec-loop-prepare-implementation-approval
Pass
Audited by Gen Agent Trust Hub on May 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions and guidance documents strictly follow a project management workflow for coding tasks. It focuses on internal project documentation and human-in-the-loop approvals.
- [DATA_EXFILTRATION]: No evidence of network activity or sensitive data harvesting was found. The skill operates locally on task files and uses relative paths within the skill's own environment.
- [REMOTE_CODE_EXECUTION]: No package installation, remote script execution, or dynamic evaluation patterns (such as eval or shell execution) were found in the skill or its examples.
- [PROMPT_INJECTION]: The content does not contain instructions to override AI safety, bypass instructions, or disclose system prompts. It uses context-setting language that is consistent with its stated purpose.
Audit Metadata