drive-discussion
Pass
Audited by Gen Agent Trust Hub on May 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides purely instructional content to guide agent behavior during design discussions. It does not contain any executable code, remote downloads, or credential harvesting logic.
- [COMMAND_EXECUTION]: The instructions encourage the use of standard codebase investigation tools such as Grep, Glob, and a 'read-only Shell'. These tools are used for diagnostic purposes to ground the agent's analysis in the actual state of the project, following a best-practice 'research before asking' approach.
- [DATA_EXPOSURE]: The skill requires the agent to read and analyze codebase content to facilitate discussions. This access is limited to the agent's authorized scope and is necessary for the skill's stated purpose of technical design and pressure-testing ideas.
- [PROMPT_INJECTION]: The skill includes instructions to 'Stay in mode until explicitly released,' which is a behavioral constraint for the agent's operational logic rather than an attempt to bypass system safety filters. No jailbreak or override patterns were found.
Audit Metadata