mobile-first
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process code from external repositories, creating a surface for potential instruction overrides if those files contain malicious content. \n
- Ingestion points: Repository instructions, layout files, and content are read by the agent as specified in SKILL.md. \n
- Boundary markers: The skill does not define specific delimiters to isolate user-provided code from the system prompt. \n
- Capability inventory: The agent can read files, write file changes (implementation), and execute commands (builds/tests). \n
- Sanitization: No specific sanitization logic is provided for the code ingested during audits. \n- [COMMAND_EXECUTION]: To verify the success of responsive repairs, the skill authorizes the execution of local build and test commands. \n
- Evidence: SKILL.md instructs the agent to 'Run the repository's build, tests, and browser checks.'
Audit Metadata