interview-me
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data including feature requests and repository facts (SKILL.md).
- Ingestion points: Data enters the agent context via feature requests and repository exploration.
- Boundary markers: The skill employs isolated role contexts and a source index to track authority (references/auto-mode.md).
- Capability inventory: The agent can write files and execute code via the to-prototype command (SKILL.md).
- Sanitization: No explicit sanitization is described.
- [DYNAMIC_EXECUTION]: The to-prototype feature allows for generating and executing code experiments to resolve design uncertainties (SKILL.md).
- [COMMAND_EXECUTION]: The skill utilizes an external runner to manage model invocations and prototype execution (references/auto-mode.md).
- [EXTERNAL_DOWNLOADS]: The skill references design patterns adapted from github.com/mattpocock/skills (SKILL.md).
Audit Metadata