verification-protocol
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for the agent to ingest and verify output from external commands such as
npm test,curl, andssh. This establishes a data processing surface for potentially untrusted tool outputs. - Ingestion points: Steps within the "Gate Function" and multiple examples in
SKILL.mdinvolve reading command-line output. - Boundary markers: Output evidence is structured within markdown code blocks.
- Capability inventory: The documentation includes examples of running test runners, build tools, and network requests via
curl. - Sanitization: No explicit sanitization or instruction-ignoring markers are defined for tool outputs.
- [COMMAND_EXECUTION]: The skill documents the use of common development commands (e.g.,
npm,curl,ssh) as part of its verification logic. These are standard tools for an agent performing development tasks and are used here within a safety-focused framework.
Audit Metadata