Quickstart

Make your repository AI-first

One page to turn any repository into a spec-driven, agent-ready codebase. Install the skill, let an agent onboard the repo, then plan and execute structured work with any coding agent.

Copy into your agent

Copy the full init.md prompt and paste it into your AI coding agent — Claude Code, Cursor, Codex, or any other. It reads the methodology and specification, installs the skill, and onboards this repository for Deep Work Plans.

https://deepworkplan.com/init.md

Self-contained agent prompt → /init.md

What this does

Adoption changes the repository in two durable ways — the pillars of the methodology.

First, the repository becomes spec-driven: work begins from a written plan and specification, not from ad-hoc prompts. Second, the repository itself becomes the agent harness — an AGENTS.md, a docs/ knowledge base, per-module docs, and a .agents/ skill home (with the .claude → .agents and .cursor → .agents symlinks) give every agent the context and commands it needs.

The adoption path

01
Verify before you install

Treat the prompt and the skill as untrusted until you have checked them. Both are open source and MIT; the skill is Markdown-first with no network calls and no telemetry. Each release publishes a SHA256SUMS over the shipped skill, so you can confirm your copy matches before running it. Releases are checksummed, not signed (signing is a documented next step).
Terminal
```
git clone https://github.com/DailybotHQ/deepworkplan-skill.git && cd deepworkplan-skill
curl -fsSL -o SHA256SUMS https://github.com/DailybotHQ/deepworkplan-skill/releases/download/vX.Y.Z/SHA256SUMS
./setup.sh --verify
```
02
Install the Deep Work Plan skill

Add the skill to your repository so any agent can plan and execute structured work. It ships a router plus nine sub-skills — create, execute, refine, resume, status, verify, onboard, author, and upgrade. Use the Skills CLI for the fastest path, or clone the repo and run setup where git and a shell are available.
Terminal
```
npx skills add DailybotHQ/deepworkplan-skill
```
or
Terminal
```
openclaw skills install deepworkplan
```
or
Terminal
```
git clone https://github.com/DailybotHQ/deepworkplan-skill.git && cd deepworkplan-skill && ./setup.sh
```
03
Onboard the repository

Invoke the onboard sub-skill and let the agent reason about the actual repo — its stack, package manager, and real validation commands. It generates AGENTS.md, a docs/ knowledge base, per-module docs, and a cross-agent .agents/ home (with the .claude → .agents and .cursor → .agents symlinks), wires thin dwp-* commands, and scaffolds a gitignored .dwp/ for plans. For large repos the onboard sub-skill uses a plan-driven path: it completes reconnaissance, then emits an onboarding Deep Work Plan. Nothing is templated; everything is adapted to your repository.
Terminal
```
/deepworkplan-onboard
```
04
Local review and opt-in addons

Onboarding installs the required AI Diff Reviewer local review (its CI gate stays optional) and offers four opt-in addons — devcontainer, Dailybot, dependency-upgrade, and design-system — that you accept only when they fit. A repo is fully conformant with zero optional addons. Use /skill-create and /agent-create (the author sub-skill) to grow stack-appropriate skills, agents, and commands beyond the baseline.
05
Plan and execute

Generate Deep Work Plans with /dwp-create and run them with /dwp-execute, then /dwp-status, /dwp-refine, /dwp-resume, and /dwp-verify as work proceeds. Each plan carries numbered tasks, validation gates, and a completion protocol — closing with a single mandatory Final Review (security pass, final-state validation, and skills reconciliation). The Executive Report remains available on request.
Terminal
```
/dwp-create <goal>
```
or
Terminal
```
/dwp-execute
```
06
Verify conformance

Run /dwp-verify for an objective pass/fail report against the specification. Confirm AGENTS.md, docs/ (with real content, not stubs), .agents/ (with thin dwp-* delegators and a catalog that matches disk), .dwp/, and tmp/ are in place — nothing templated, everything reasoned for this repo.
Terminal
```
/dwp-verify
```

The outcome

The repository becomes spec-driven and agent-pilotable: the plan is the durable source of truth, and the repository itself becomes the harness any agent runs against.

Next steps

Read the methodology Browse the spec Browse the kit See examples

Verify before you install

Install the Deep Work Plan skill

Onboard the repository

Local review and opt-in addons

Plan and execute

Verify conformance