workflow-e2e

Pass

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill references a local Node.js script (.claude/scripts/lib/workflow-skill-contract.cjs) to synchronize workflow contracts. This is a legitimate maintenance action for the skill's state management within a development environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data sources to automate test generation.
  • Ingestion points: Processes browser recording JSON files, UI diffs, and product source code as specified in SKILL.md under the Source Dispatch section.
  • Boundary markers: Implements structured verification gates such as tests-pass, Round Integrity Check, and Case traceability to ensure output validity.
  • Capability inventory: Performs file system writes (authored tests), spawns sub-agents for specialized tasks, and executes E2E runner commands.
  • Sanitization: Employs a bounded convergence loop (--fix-loop) and manual visual review passes ($experience-review) to verify generated content against project requirements.
  • [SAFE]: The protocol follows standard development practices, including least-privilege sub-agent usage and explicit triage steps to manage risk. No external downloads from untrusted sources or credential exfiltration patterns were found.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 4, 2026, 10:12 AM