Top AI Product

Give an AI agent a vague task and it drifts; give it the same task with a real plan and it behaves like your best engineer. Deep Work Plan, an open-source project born at DailyBot, is built on that idea — it turns any repository into a harness that keeps agents on track.

## Plan as the source of truth

Before any code, you write a spec: a goal, atomic tasks, and for each task explicit acceptance criteria plus a validation gate. “Done” is decided by the gate, not by how the model feels it went. Deep Work Plan installs that harness into the repo itself — an AGENTS.md, docs, an .agents/ skills home, and the DWP skill — so any agent can pilot any repo and judge its own work against the same bar.

## Built to survive long runs

The other half is durability. The plan lives in the repo with atomic tasks and resumable state on disk, so long runs survive context resets — any agent picks up exactly where the last one left off. It is MIT-licensed and works with Claude Code, Codex, Cursor, or whatever agent comes next. The pitch is blunt: models matter, but context matters more, and a written plan is how you give an agent both.