Part

Agent-ready codebase

You might say: agent experience, AX, agent legibility, harness engineering, the playground

The repository set up so an agent can find its way, make a change and prove the change works with no person watching. It holds a short instructions map, a glossary of the project's terms, one verify command, a tool that drives the running app, the tools and skills the agent may call, and checks that tighten after each repeated mistake.

What it does

Lets the same agent and model do good work that it can prove, so less of its output needs a person to read it.

Ask

Can a fresh agent find the right code, run the one verify command, drive the app and show evidence, all without asking a person?

What to look for

1. A short AGENTS.md map and a glossary near the repository root
2. One verify command that runs the same checks as CI
3. A checked-in CLI or script that starts and drives the app
4. Lint rules and checks that cite the review comment they replaced

Parts

• Codebase health: How easy the code is to find and change.
• Instructions: The durable guidance an agent receives about the repository, task boundaries, checks, style, and safety.
• Shared vocabulary: A short file of the project's terms, each with one meaning, that agents and people both use.
• Verify command: One command that runs every check the repository requires, the same way on a laptop, on a runner and in CI.
• Control CLI: A checked-in command-line tool that starts the app, signs in and drives its features.
• Tool access: The tools and skills an agent may call during a run.
• Ratchet: A stricter check is switched on while known failures remain in an exception list.

Connections

• Raises: Merge rate
• See also: Verification

Sources

1. Matt Pocock, AI Coding Dictionary, AXhow well the environment is set up for an agent to do good work in a codebase.
2. Matt Pocock, AI Coding Dictionary, AXWhen the same agent performs well in one repo and badly in another … same model, same harness … the difference is usually AX.
3. Matt Pocock, AI Coding Dictionary, AXThe instinct is to blame the model or rewrite the prompt; the fix is more often in the repo.
4. Kent C. Dodds, kcd-skills, autopilot skillAudit any repository and set it up so coding agents ship changes safely with little human review.
5. Kent C. Dodds, kcd-skills, autopilot skillThe repo tells agents what to build and how to prove it.
6. Kent C. Dodds, You Own the Outcome: Product Engineering in the Age of AI, Design the playgroundBuild good primitives. Agents are only as good as the building blocks you give them.
7. Kody, kody-hq/kody repository, docs/contributing/harness-engineering.mdHumans steer outcomes; agents execute implementation details.
8. Lauren Tan, pstack, Verify the result and open a PRMake the agent able to do it, and you stop being the bottleneck.

Use it on your project

teardown-apply: hold "Agent-ready codebase" (agent-ready-codebase) from the teardown at https://teardownatlas.com/software-factory/ against my project

Paste it into a coding agent that has the teardown skills.

An AI agent wrote this map with the teardown skill. It can contain errors: check the sources before you rely on a part. About and legal