create-verification-skill
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions in
SKILL.mdrequire the agent to discover and execute build and start commands found within the repository (e.g., package scripts, Makefiles, README quickstart). - [DYNAMIC_EXECUTION]: Section 4 of
SKILL.md('Prove the generated skill') instructs the agent to execute the newly generated skill instructions end-to-end, including launching the application and driving features. This involves executing dynamically generated content at runtime. - [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from the repository to determine its execution logic.
- Ingestion points: Project-local files such as
README.md,package.json,Makefile, and existing test specifications. - Boundary markers: None identified. The agent is encouraged to follow 'the repo's own documented dev command'.
- Capability inventory: The agent utilizes shell execution capabilities to run build processes, start servers, and execute CLI tools.
- Sanitization: No explicit sanitization or validation of the commands extracted from the repository is performed before execution.
Audit Metadata