ai-regression-testing

Pass

Audited by Gen Agent Trust Hub on Aug 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: Documentation and code snippets for regression testing using Vitest and Next.js follow industry standards for software verification.
  • [SAFE]: The test setup isolates the environment by clearing Supabase keys (e.g., NEXT_PUBLIC_SUPABASE_URL), reducing the risk of accidental production data access during testing.
  • [COMMAND_EXECUTION]: The skill facilitates the execution of local build and test scripts (npm run test, npm run build). These commands are standard development operations and are restricted to the local project environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow where the agent processes output from shell commands to verify code integrity. 1. Ingestion points: Output from local build and test processes. 2. Boundary markers: The skill uses a structured three-step workflow (Test, Build, Review) to separate command execution from AI analysis. 3. Capability inventory: The agent is limited to local shell command execution for testing. 4. Sanitization: The agent is expected to interpret raw output to identify regressions, which is a benign use of this capability for the skill's stated purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 15, 2026, 03:24 AM
Security Audit — agent-trust-hub — ai-regression-testing