red/green TDD

Overview

red/green TDD is a test-driven development loop adapted for coding agents: first generate tests, run them to confirm failure (red), implement code to satisfy those tests, and then rerun to confirm success (green). In the agentic coding context, this pattern turns a vague implementation request into a structured workflow with explicit checkpoints, making autonomous code generation more reliable and easier to validate.

For AI Product Managers, red/green TDD matters because it offers a practical control mechanism for AI-assisted software delivery. Instead of asking an agent to "build the feature" end-to-end, teams can constrain work into small verifiable iterations. That improves traceability, reduces silent failures, and creates clearer acceptance criteria for agent-generated output—especially as tools like Claude Code and GPT-5.4 make autonomous coding more common in product development workflows.

Key Developments

• 2026-02-23: Simon Willison’s Agentic Engineering Patterns was referenced as a hub for building and operating agentic systems, including red/green TDD as a specific pattern for coding agents.
• 2026-04-03: A detailed description of the prompt pattern appeared: invoking “red/green TDD” tells agents to write tests first, run them to verify failure, implement code, and rerun tests to verify success.
• 2026-04-04: Lenny Rachitsky shared Simon Willison’s view that late 2025 marked an inflection point for AI coding, with autonomous coding agents, benchmark progress, and “thin templates” helping drive adoption of red/green TDD.

Relevance to AI PMs

• Define measurable agent workflows: PMs can translate feature requirements into test-first tasks, giving coding agents explicit success criteria and reducing ambiguity in implementation.
• Improve reliability in AI-assisted delivery: By requiring a red-to-green loop, teams can catch failures earlier and prevent agents from producing plausible but unverified code changes.
• Operationalize agent governance: Red/green TDD can be built into product engineering processes, CI pipelines, and acceptance reviews so AI-generated code is evaluated against observable outcomes rather than narrative confidence.

Related

• Simon Willison: A key advocate for agentic engineering patterns and an early source framing red/green TDD for coding agents.
• Lenny Rachitsky: Amplified the idea through commentary on the AI coding inflection point and its implications for software teams.
• Claude Code: An example of a coding agent environment where red/green TDD can structure autonomous implementation work.
• GPT-5.4: Cited as part of the model progress that made instruction-following strong enough for these coding workflows to become more dependable.
• agentic-engineering-patterns: The broader pattern library in which red/green TDD is positioned as one tactic for reliable agent operation.
• test-driven development / red green test-driven development / red-green TDD: Common aliases and adjacent terminology connecting this concept back to traditional TDD practices.