The methodology site for planning and executing complex software work with AI coding agents.
deepworkplan.com is the official website for the Deep Work Plan (DWP) methodology — a structured, agent-agnostic approach to executing complex, multi-step software work reliably with AI coding agents.
Models matter; context matters more. DWP turns any repository into a structured environment — context, guardrails, and a durable plan — where any coding agent executes with precision and finishes long-horizon work. Put another way: Deep Work Plan is spec-driven development where the repository itself becomes the harness.
The methodology rests on five ideas, and the site is built to argue them:
- You steer; the agents do the hours. You decide what done means and where the lines are. The plan carries your intent, so the work does not need correcting every twenty minutes.
- The plan is what the agent returns to. Long work fills any model's context and detail falls away. Atomic tasks, validation gates, and resumable state give the agent something durable to come back to.
- Done is a contract, not a feeling. Every task names its acceptance criteria and the checks that must pass. An agent does not get to decide it finished — it passes, or the task stays open.
- The repository is the harness. Context, tools, guardrails, and state live in the repo as plain files any agent can read. No lock-in, no external brain; it survives a context reset or a change of agent mid-flight.
- Context is the scarcest resource. Instructions load progressively by trigger, validation is selected from what each task touched, and skills are decided task-locally — so the plan pays for itself.
Work is planned at the size it deserves: a compact Lite plan for bounded changes, a per-task Full plan for long-horizon work, with the same contract — stable task ids, touched surface, acceptance criteria, gates, completion evidence, and one final review — behind both.
The site is a fast, fully multilingual static site built with Astro — 17 languages ship today (English, Spanish, Portuguese, Chinese, Japanese, German, French, Korean, Russian, Italian, Turkish, Indonesian, Vietnamese, Hindi, Polish, Ukrainian, Thai). It explains and sells the methodology, hosts the readable specification, and catalogs the kit (presets, adapters, commands) for installing DWP into any repository. The repository dogfoods the methodology it documents.
The repository practices the methodology it documents. The official DeepWorkPlan skill is vendored under .agents/skills/deepworkplan/ — the whole pack is committed and pinned by skills-lock.json, so every agent that clones this repo gets the exact same harness with no install step. Alongside it live the thin dwp-* command delegators, the /skill-create and /agent-create delegators (which route to the skill's author sub-skill that grows the repo's own kit), the /lib-upgrade delegator for the opt-in dependency-upgrade add-on, and the /init adoption flow. All Deep Work Plan output lives in the gitignored .dwp/ directory. Running /dwp-verify here checks this repository against the same standard the site publishes. See docs/ARCHITECTURE.md for details.
| Feature | Description |
|---|---|
| 📚 Methodology reader | The DWP methodology, principles, and adoption workflow |
| 📐 Specification reader | The normative DWP standard: task anatomy, validation gates, archetypes |
| 🧰 Kit catalog | Presets, adapters, and commands for installing DWP into a repo |
| 🌐 Multilingual | 17 languages (en, es, pt, zh, ja, de, fr, ko, ru, it, tr, id, vi, hi, pl, uk, th) with route parity |
| 🌙 Dark mode | Inline hurricane-lamp toggle in the masthead with localStorage persistence |
| ⚡ Performance-first | Static site, minimal JS, optimized assets, 90+ Lighthouse |
| 🔍 SEO + AEO | Sitemap, Open Graph, structured data, agent-friendly Markdown endpoints |
🚀 /init adoption endpoint |
A canonical, copy-paste onboarding prompt at deepworkplan.com/init — paste it into any agent and it adopts the methodology in the current repo |
| 🔌 Agent API + MCP | Markdown endpoints for every page, a published openapi.json, and an MCP server at the edge so agents can query the methodology directly |
| 🤖 Agent-friendly | Shared .agents/ home for skills, commands, and agent definitions across Claude Code, Codex, Cursor, Gemini |
Note: The site focuses on the methodology, specification, and kit readers plus agent-friendly Markdown endpoints. Earlier blog, slides/tech-talks, and personal pages have been removed.
| Layer | Technology |
|---|---|
| Framework | Astro 7.x |
| UI | Svelte 5.x |
| Styling | Tailwind CSS 4.x |
| Language | TypeScript 6.x |
| Linting | Biome 2.x |
| Testing | Vitest + Testing Library |
| Content | Markdown, MDX |
| Edge | Cloudflare Pages Functions (bot analytics middleware) |
| Hosting | Cloudflare Pages |
Requires Node.js 24+ (CI runs on 24.20.0). The package manager is pnpm, pinned via Corepack in
package.json#packageManager. Enable it once withcorepack enable; the dev container does this automatically.
corepack enable
pnpm install
pnpm run devVisit http://localhost:5555 to preview.
| Command | Description |
|---|---|
pnpm run dev |
Start dev server at localhost:5555 |
pnpm run build |
Production build (prebuild regenerates .agents/skills/index.json, then astro check && astro build) |
pnpm run astro:preview |
Preview production build locally |
pnpm run biome:check |
Lint and format check |
pnpm run biome:fix |
Auto-fix lint and format issues |
pnpm run astro:check |
TypeScript type checking |
pnpm run test |
Run unit tests (Vitest) |
pnpm run test:watch |
Vitest in watch mode |
pnpm run test:coverage |
Tests with coverage report |
pnpm run images:optimize |
Convert staged images to WebP |
pnpm run md:check |
Verify HTML / Markdown agent-endpoint parity |
pnpm run md:check:strict |
Same as md:check; exits 1 on missing (for CI) |
pnpm run i18n:check |
Verify translation parity across all 17 active languages |
pnpm run i18n:scaffold |
Scaffold strings + content for a new language code |
pnpm run lighthouse |
Run Lighthouse CI audit (mobile) |
pnpm run lighthouse:desktop |
Run Lighthouse CI audit (desktop) |
pnpm run ncu:check |
Check for dependency updates |
pnpm run release |
Bump version and create release commit |
├── public/ # Static assets (images, fonts, icons, robots.txt)
├── src/
│ ├── components/ # Astro & Svelte components
│ ├── content/ # Content Collections (17 language folders per collection)
│ │ ├── methodology/ # Methodology docs by language (en/, es/, pt/, zh/, …) — primary content
│ │ ├── spec/ # Specification reader content (en/, es/, pt/, zh/, …)
│ │ ├── kit/ # Kit catalog: presets, adapters, commands (en/, es/, pt/, zh/, …)
│ │ └── pages/ # Agent-friendly Markdown endpoints (en/, es/, pt/, zh/, …)
│ ├── layouts/ # MainLayout, InternalLayout, ...
│ ├── lib/ # Utilities, types, translations (one `<code>.ts` per active language)
│ ├── pages/ # File-based routing (EN at root, non-EN under /[lang]/ dynamic tree)
│ ├── middleware.ts # Route allowlist & rewrites (derived from the language registry)
│ └── styles/ # Global CSS (Tailwind 4)
├── functions/ # Cloudflare Pages Functions (edge middleware)
├── tests/ # Vitest unit & integration tests
├── scripts/ # Build utilities (image optimization, parity checks)
├── docs/ # Project documentation
├── .agents/ # Cross-agent skills, commands, agent definitions
├── .claude → .agents # Backward-compat symlink for Claude Code
├── assets/ # README and documentation assets
└── astro.config.mjs
| Document | Description |
|---|---|
| Product Spec | Vision, audience, features, and content strategy |
| Architecture | Technical implementation and patterns |
| Development Commands | Build scripts and workflows |
| Standards | Coding conventions |
| I18N Guide | Multilingual content and translations |
| Testing | Vitest setup and test conventions |
| Performance | Astro SSG optimization and budgets |
| Accessibility | WCAG AA standards and patterns |
| SEO | Meta, structured data, multilingual, AEO |
| Security | Security best practices |
| AI Agent Onboarding | Setup for AI coding assistants |
| Skills & Agents Catalog | Available skills and agents |
Deployed to Cloudflare Pages on every push to main:
- Cloudflare triggers a build
- Runs
pnpm run build(regenerates the agent-skills index, thenastro check && astro build) - Serves the
dist/folder from its global CDN
Live at deepworkplan.com.
MIT — Dailybot, Inc.
- Icons: Heroicons, Simple Icons
🔌 Powered by Dailybot
Dailybot is an AI-powered async communication platform that keeps people and agents visible — without adding more meetings or tools. It lives where your team already works (Slack, Teams, Google Chat, Discord, VS Code, and the CLI) and turns scattered signals into clear progress: async check-ins and standups, AI summaries that detect blockers and read team sentiment, workflow automation and approvals, team analytics, and recognition. As AI agents join the workflow, Dailybot surfaces their status and activity right alongside your team's — so long-running agents never go dark. Learn more.