mushi-integration
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external websites to generate user stories and test cases, creating an attack surface for indirect prompt injection.
- Ingestion points: Stage 3 (
mushi stories map --url https://your-app.com) crawls external content. - Boundary markers: Absent in the skill instructions; it relies on the underlying LLM's handling of crawled data.
- Capability inventory: The skill enables SQL queries via MCP, CLI command execution, and local script execution (Playwright tests).
- Sanitization: The instructions recommend an 'approval queue' and manual review (
mushi tdd approve) before executing generated tests, which mitigates but does not eliminate the risk. - [DYNAMIC_EXECUTION]: The skill automates the generation and execution of code based on AI-triaged reports and crawled stories.
- Evidence: Stage 4 (
mushi tdd gen) and Stage 5 (mushi tdd run) generate and execute dynamic Playwright scripts. - Context: This behavior is aligned with the skill's purpose as a TDD smoke test, and the presence of a manual approval command suggests a controlled workflow.
Audit Metadata