e2e
Pass
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted input from feature files or scenario descriptions without sanitization or boundary markers. 1. Ingestion points: Feature file paths or scenario descriptions provided as arguments. 2. Boundary markers: Absent; the input is interpolated directly into prompts for sub-agents. 3. Capability inventory: Sub-agents can write files and execute commands. 4. Sanitization: Absent.
- [COMMAND_EXECUTION]: The skill involves the dynamic generation and execution of TypeScript code. Execution method: The playwright-test-healer sub-agent is instructed to run all generated tests. Source of code: Test scripts are dynamically produced by an LLM-based generator, posing a risk of executing unintended logic if the source scenarios are malicious.
Audit Metadata