test-qa
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from untrusted sources, including the application's DOM, network logs, and codebase (ingestion points in Phase 0 and Phase 2). The lack of boundary markers and sanitization, combined with the agent's ability to interact with the browser and database (capabilities inventory: playwright-cli, supabase:execute_sql), creates a risk where malicious input could influence agent actions.
- [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill uses
npx --yes @playwright/cli@latestto download and run browser automation tools at runtime from a remote registry. - [DATA_EXPOSURE_AND_EXFILTRATION]: The agent is directed to scan for and read environment files such as
.env.localand.env.testto extract credentials. It also usessupabase:execute_sqlto interact with the backend database. - [COMMAND_EXECUTION]: The skill performs shell commands to identify running processes and executes browser automation commands to simulate user behavior.
Audit Metadata