flow-qa-task-runner

Pass

Audited by Gen Agent Trust Hub on Oct 31, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from external specification files to drive test logic and shell execution.
  • Ingestion points: The agent is instructed to read qa_session.json and segments of PRD/techspec files in Phase 1 to define the scope and parameters of execution.
  • Boundary markers: There are no explicit delimiters or specific warnings to ignore instructions embedded within these external specification files during processing.
  • Capability inventory: High-privilege capabilities including the execution of curl, docker, and playwright-cli are triggered based on the content of these files.
  • Sanitization: No input validation or sanitization is defined for the content extracted from external files before it is interpolated into shell commands or generated scripts.
  • [COMMAND_EXECUTION]: The skill is designed to construct and execute shell commands including curl for API testing, docker run for database validation, and playwright-cli for UI testing.
  • [EXTERNAL_DOWNLOADS]: Fetches official database images (e.g., postgres:16-alpine, mysql:8, mongo:7) from well-known registries to perform database validation tasks.
  • [DYNAMIC_EXECUTION]: Dynamically generates shell scripts and Playwright TypeScript test files (qa_task_[N]_[slug].spec.ts) at runtime based on the provided task definitions and specifications.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 31, 2026, 01:05 PM