AI-Driven E2E Testing

A five-phase workflow that uses AI to drive end-to-end testing across multi-framework monorepos, turning testing from a manual bottleneck into an agent-assisted pipeline.

What it is

Viking (@vikingmute) developed a reproducible workflow for AI-assisted E2E testing in TinyShip, a monorepo containing Next.js, Nuxt.js, and TanStack Start with both PostgreSQL and SQLite support. Each feature change has six possible combination outcomes, making manual testing impossible.

The workflow has five phases: Spec → Code → Verify → Test → Green.

Why it matters

Multi-framework monorepos multiply test complexity combinatorially. A single feature change can affect multiple frameworks, databases, and deployment targets. Traditional manual QA cannot cover the explosion of combinations. AI-driven E2E testing treats the test matrix as a structured problem that agents can systematically explore.

Key points

• Spec phase: Define the feature behavior in natural language or structured format before writing code.

• Code phase: Implement the feature across all affected frameworks.

• Verify phase: Use agent-assisted verification to confirm the implementation matches the spec.

• Test phase: Run E2E tests across all framework-database combinations.

• Green phase: All combinations pass before merge.

• Combinatorial explosion is the core challenge: TinyShip's six combinations per feature change represent a class of problems that grow faster than manual QA capacity.

• Template-driven new feature development: The workflow provides a concrete template for E2E-driven feature development with AI assistance, not just test execution.

• Agent-browser as QA replacement: For multi-framework repos, agent-browser tools can replace manual QA by systematically navigating applications and verifying behavior.

Open questions

• Does this workflow generalize beyond monorepos to microservices or distributed systems?
• How does the agent handle flaky tests or non-deterministic UI behavior?
• What is the cost trade-off between agent-driven E2E testing and traditional automated test suites?

Prompts for witness

• Where in your current projects is manual testing the bottleneck? Would a Spec→Code→Verify→Test→Green workflow help?
• If you had to test six combinations for every feature change, what would you automate first?

Related

• Agent-Driven QA — autonomous QA agent pattern
• Self-Verification Loops — verification as harness component
• Claude Code Overview — agent-assisted development workflows