Deep Work Plan
FeaturedTurn any ambiguous goal into a plan every AI agent can execute.
Deep Work Plan is a full agent methodology installed as a skill pack. Give the agent a fuzzy multi-step goal — "rebuild our developer portal", "migrate the payments service", "add SAML to the auth stack" — and it produces a numbered, atomic, validated plan under `.dwp/plans/`. Each task has an explicit validation gate. The agent runs one task at a time and refuses to move on until validation passes. No lost progress if the session ends. No half-done tasks masquerading as done. No missed steps.
8
sub-skills covering the full plan lifecycle
5+
AI coding agents supported
0
silent failures — every gate is explicit
MIT
licensed, end to end
Sub-skills
One skill pack. Every verb your agent needs.
Each sub-skill maps 1:1 to a public Dailybot API endpoint. Composable, coherent, and documented — the way your agent expects.
create
Decompose a fuzzy multi-step goal into a numbered, gated plan. The agent asks clarifying questions if the goal is under-specified.
execute
Run the plan task-by-task, validating every gate. The agent refuses to advance if validation fails, so 'done' means actually done.
refine
Add, remove, or reorder tasks mid-flight without losing completed work. Refine keeps state, adjusts scope, and re-runs from the right point.
resume
Reconstruct plan state and continue an interrupted run. Session crashes, laptop closes, agent gets swapped — the plan picks up where it left off.
status
Read-only progress report — completed, in-progress, blocked. Perfect for the standup, code review, or a quick stakeholder ping.
verify
CONFORMANT / NOT CONFORMANT audit against the plan spec. Objective, boolean, and reproducible — a plan either matches its own contract or it doesn't.
onboard
Re-run onboarding on an existing repo. Reconciles, never clobbers — safe to run on a repo that already has DWP set up.
author
Author or update the repo's skills, agents, and commands. The methodology grows with your workflow.
Why it works
The bits that make it feel like a teammate.
Atomic, numbered tasks
Each task ships one deliverable. No 'build the whole feature' catch-alls. The plan is a numbered checklist an agent can execute deterministically, in order, without ambiguity.
Explicit validation gates
Every task carries an explicit criterion for 'done' — a passing test, a green build, a specific file existing, a curl returning 200. The agent refuses to advance until the gate passes.
State that survives sessions
All plan state lives in `.dwp/plans/PLAN_{name}/state.json`. If the agent crashes, the laptop reboots, or you swap Cursor for Claude Code, the plan resumes from the exact task it stopped on.
Refinable without losing work
Add a task in the middle, remove a task that turned out irrelevant, reorder for a better dependency graph — `refine` preserves everything already completed. Scope changes stop being destructive.
Standup-wired via the Dailybot addon
Kickoff, mid-plan progress, blockers, and completion all report to Dailybot automatically as first-class updates. Your team follows the plan in the same feed they already read every morning — no separate DWP dashboard to check.
Every AI CLI, same methodology
Claude Code, Cursor, OpenAI Codex, Gemini CLI, Copilot — same skill install, same plan structure, same gates. Swap agents mid-plan and nothing about the methodology changes.
How Dailybot connects
Kickoff, mid-plan progress, blockers, and completion all report to Dailybot as first-class updates. Your team watches the plan advance in the same feed they already read every morning.
Built for
Compatible agents
Install
Two commands and your agent is a teammate.
Install the Dailybot CLI once — it authenticates your session and every skill calls the same binary underneath. Then add the skill to whichever AI CLI you use.
Step 1
Install the CLI (recommended baseline)
One command. Sets up the `dailybot` binary and prepares your local auth.
$ curl -sSL https://cli.dailybot.com/install.sh | bashStep 2
Install in your AI CLI of choice
Every major AI CLI reads skills from your project's `.agents/skills/` folder. Pick your CLI, run the command, and the skill is available in every future session.
curl -sSL https://cli.dailybot.com/install.sh | bash
npx skills add DailybotHQ/deepworkplan-skillnpx skills add DailybotHQ/deepworkplan-skillgit clone https://github.com/DailybotHQ/deepworkplan-skill \
.agents/skills/deepworkplangit clone https://github.com/DailybotHQ/deepworkplan-skill
cd deepworkplan-skill && ./setup.shIn practice
What teams actually do with it.
01
Rebuild the developer portal
One fuzzy goal, one numbered plan. Twelve atomic tasks (schema, content collection, page component, routes, mirrors, redirects, tests). Each with its own gate. The agent ships one PR-sized unit at a time — reviewable, revertable, resumable.
02
Migrate the payments service
The kind of work that used to sprawl for weeks with no clear 'am I 30% or 80% done?' answer. DWP produces the plan, and the human always knows exactly which tasks are complete, which are blocked, and which are the critical path.
03
Refactor across many files without losing the plot
Twenty-plus files touched, each with a small change. DWP's numbered tasks and validation gates make a big refactor feel like small refactors done in sequence — no half-migrated state, no orphaned commits.
04
Onboard a new agent to an existing repo
`dwp-onboard` runs the same setup a human would do — writes AGENTS.md sections, adds skills, wires hooks — and reconciles with what's already there. Safe to run on a repo that already has DWP configured.
Our team stopped writing multi-week 'design docs that never actually match what we ship'. Now the plan is the contract — one Markdown file, twelve tasks, each with its gate. The agent runs the plan and we always know where we are.
Staff engineeron team using DWP for migrations + refactors
FAQ
The questions we hear the most.
Where do plans live?
`.dwp/plans/PLAN_{name}/` — a directory per plan, with tasks, state, and progress. It's a normal folder of Markdown and JSON, so any editor or grep works on it.
What happens if a task fails its validation gate?
The agent stops, reports the failure to Dailybot, and refuses to move on. Fixing the failure is a real, visible task — not a silent skip. This is what makes 'done' actually mean done.
Can I refine a plan mid-execution?
Yes. `refine` lets you add, remove, or reorder tasks without losing completed work. Scope changes are safe — the plan preserves state and re-runs from the right task.
How does DWP interact with the Dailybot Agent Skill?
DWP ships an optional Dailybot addon that reports plan lifecycle events (kickoff, significant task, blocked, completion) via the same Agent Skill your team already has installed. You can also disable the addon if you don't use Dailybot.
Do I need to author plans by hand?
No. The `create` sub-skill decomposes a fuzzy goal into a numbered plan for you. If the goal is under-specified, the agent asks clarifying questions before drafting.
What if the session dies mid-plan?
Run `dwp-resume`. It reconstructs plan state from `.dwp/plans/PLAN_{name}/state.json` and picks up on the exact task it stopped on. No lost progress.
What it is
Deep Work Plan is a full agent methodology packaged as a skill. Where the Dailybot Agent Skill teaches your agent to talk to your team, DWP teaches your agent to ship multi-step work without losing the plot. It’s the same methodology we use internally at Dailybot to run large refactors, migrations, and portal rebuilds.
Why teams use it
Large multi-step work is where agents historically break down. They start strong, get 40% through, and then either loop, forget the plan, or declare victory prematurely. DWP fixes all three: numbered atomic tasks, explicit validation gates, and durable state that survives sessions. The result is agent work that reads like a human’s work — small, verifiable, and always resumable.
How it works with Dailybot
DWP is standalone — the methodology and skill pack work with any AI coding agent. The optional Dailybot addon reports plan kickoff, mid-plan progress, blockers, and completion as first-class Dailybot updates. Your team follows the plan in the same feed they already read every morning, without leaving Slack, Microsoft Teams, Discord, or Google Chat.
Ship agent work your team can actually see.
Install the skill in your favorite AI CLI and every action your agent takes lands in the same standup your humans already read.