create-verification-skill
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies verification steps by analyzing content from the target repository's codebase, which could contain untrusted instructions.
- Ingestion points: The agent reads the local repository's README, package scripts, and source code (SKILL.md Section 1).
- Boundary markers: No explicit markers are defined for delimiting repo data during the interview process.
- Capability inventory: The skill utilizes terminal access for executing shell commands, browser automation tools (preview_*), and file system writes to generate local skills and capture evidence.
- Sanitization: The instructions explicitly mandate environment isolation (unique ports and data directories) and side-effect verification to ensure commands behave as expected.
- [DYNAMIC_EXECUTION]: The skill is designed to generate a local skill configuration (.agents/skills/verify-/SKILL.md) and then immediately execute it to prove functionality. This involves script generation and the execution of shell commands derived from the project's build and test infrastructure.
Audit Metadata