flow-runner

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands via npx @playwright/cli. This is a standard method for running the Playwright automation framework and is documented as the preferred driver for token efficiency and artifact persistence.
  • [CREDENTIALS_UNSAFE]: The instructions include security best practices for credential management. It directs the agent to check environment variables silently (printenv NAME >/dev/null) to avoid printing secrets to the log and suggests using saved session states (storageState) to prevent password leaks in the command transcript.
  • [PROMPT_INJECTION]: Detection of Indirect Prompt Injection surface: 1. Ingestion points: accessibility trees and snapshots from external URLs navigated to during testing. 2. Boundary markers: No explicit instructions provided to the agent to treat page content as untrusted data or use delimiters. 3. Capability inventory: Browser navigation, clicking, typing, shell execution (npx), and file writing. 4. Sanitization: No mentioned sanitization or filtering of external content before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 04:46 PM
Security Audit — agent-trust-hub — flow-runner