ps-create-verification-skill
Warn
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill requires the agent to generate and then execute project-specific shell commands for launching and interacting with applications.
- Evidence:
SKILL.mdStep 4 ("Prove the generated skill before handing it over") instructs the agent to execute the instructions it just generated in Step 2 ("Generate the skill"). - The generated commands for "Launch", "Doctor", "Drive", and "Cleanup" are derived from the local codebase, which could lead to execution of unexpected or harmful commands if the codebase is untrusted.
- [INDIRECT_PROMPT_INJECTION]: The process of "interviewing" a repository to generate instructions creates a surface for indirect prompt injection.
- Ingestion points: Repository content including
README.md,package.json,Makefile, and existing test specifications. - Boundary markers: None. The skill does not provide delimiters or instructions to ignore potential commands embedded in the codebase.
- Capability inventory: The agent is granted the ability to run shell commands, interact with the filesystem, and perform network requests (browser actions) to verify the application.
- Sanitization: There are no instructions to sanitize or validate the commands or parameters extracted from the codebase before execution.
Audit Metadata