Claude Code
Setup guide and native plugin format are ready, so tasks can be claimed and handed off over MCP. We have not completed an install check in Claude Code yet; feedback welcome.
Claude Setup Guide THE SHARED TASK LAYER FOR AI AGENTS
Agents come and go. Tasks remain.
One task list shared by ChatGPT, Claude Code, Cursor, and every agent you use. Switch tools, keep the context.
Plan → Code
01 / 04
SD-101
Stripe Webhook
PLAN
Add Stripe webhook & retry queue!
YouDeveloper
ChatGPTPlanner
StateDockIDLECodexBuilder
It records state; it doesn't plan or run the work for you. A one-click local version and self-hosting are available. One person, one agent, one repo? A TODO.md is enough.
REAL INTERFACE · PRODUCT SHOWCASE
Clean, focused, and built for real engineering tasks.
No cluttered chat transcripts. A structured shared task layer with clear states, team handoffs, and instant status updates.
Cross-project monitoring and agent status
Unified view of active projects, latest task handoffs, and connected agent activity.
MULTI-AGENT & TOOL ECOSYSTEM
Plan on the web, code in the terminal, on the same task.
Connect ChatGPT web through OAuth 2.1, and use Codex, Antigravity, DeepSeek Harness, Claude Code or Cursor from terminals and IDEs. Different agents share one task list, so context carries over.
Community
Cursor (IDE)
Follow the docs to set up .cursor/mcp.json and a rules file to read tasks and write handoffs. We have not completed an install check in Cursor yet.
Cursor SetupOpenAI Codex
Most of StateDock's own tasks are claimed and handed off by Codex over MCP; the plugin passed an isolated Codex CLI install check.
Codex SetupChatGPT Web & App
Connect StateDock Remote MCP to ChatGPT web through OAuth, then refine requirements and save them as tasks.
OpenAI MCP GuideGoogle Gemini Web
Uses standard Remote MCP and OAuth, so it should connect; we have not verified it ourselves yet. Reports welcome.
Gemini SetupGoogle Antigravity
Many of StateDock's own tasks are delivered by Antigravity over MCP. A native plugin adapter is built; in-app OAuth and Marketplace listing are not verified yet.
Skill SpecDeepSeek Harness
Published on npm (@statedock/dsh-plugin). CLI and Skills are verified; desktop GUI and model-run acceptance is pending. Authentication uses an environment token.
npm @statedock/dsh-pluginChrome Extension (Quick-Panel)
A toolbar badge shows waiting decisions; capture a note from any page and answer waiting tasks in the panel.
Extension DetailsOpen Protocol & Custom Agents
Remote MCP uses OAuth 2.1 with PKCE and publishes RFC 8414 / RFC 9728 discovery metadata. Any MCP client can connect, and task results follow a tool-neutral public record format.
Task Record Spec v1Stable: used daily to build StateDock itself. Beta: published and partly verified. Community: docs or adapters exist, but no install check in that product yet.
WHY STATEDOCK · ARCHITECTURE COMPARISON
Why not just a TODO.md or GitHub Issues?
They work, until you have more than one agent, device, or person. Here's where StateDock fits.
| Capability | TODO.md / AGENTS.md | GitHub Issues / Linear + MCP | StateDock (Shared Task Layer) |
|---|---|---|---|
| Best for | Solo dev, single agent, one repo | Teams with existing sprint workflows | Solo dev or small teams with multiple agents |
| Web Planning (ChatGPT) | Manual copy-paste required | Depends on custom MCP connectors | OAuth 2.1 connection, saves directly to Ready queue |
| Agent task claiming | Informal markdown comments, conflict-prone | Requires custom scripts and label flows | Native Ready queue, assigned by agent identity |
| Handoff records | Unstructured free-form text | Scattered across issue comments | Structured 3-part handoff (changes/verification/notes) |
| Mobile progress & decisions | None, must stay at your terminal | Generic alerts, not tailored to agent blockers | Push alerts to Discord, Telegram, Bark & one-tap approval |
| Data ownership & deployment | Your Git repository | Third-party commercial SaaS | Open source self-hosting (Docker/CF) or Cloud |
UNIFIED NOTIFICATIONS
No window juggling. Know what's happening at a glance.
One agent fixing your web app, another building mobile features. Lost track of what finished where? Let project updates come to you.
Project names, readable task numbers, and status tags travel with each update. Started, completed, or blocked — check it on your phone.
DiscordTelegramBarkntfyCustom Webhooks
Connect your preferred channels in Settings and filter projects/events. Notifications alert you on completions and blockers; custom webhooks integrate with your own tools.
Your Task FeedSample feed · Start notifications enabled
Task Started
Fix mobile image upload handling
Task Completed
CSV export supported; verification suite passing.
Needs Your Decision
Should legacy image formats also be converted? Please confirm scope.
OPEN-SOURCE LOCAL & SELF-HOSTINGAGPL-3.0-only
View sourceOpen source. Your own instance.
Run StateDock on your laptop, server, or cloud. StateDock Local (one program for your computer, a Linux server, or Docker) and the Cloudflare edition are fully open source.
Prefer a fully managed hosted service? Use StateDock Cloud.Open StateDock Cloud
A REAL PROJECT · PUBLIC READ-ONLY DEMO
See how StateDock is built.
Explore the goals, decisions, and agent outcomes recorded in StateDock’s English project mirror. See the real development record, not a blank sample.
Explore the full projectStateDock-devREAD ONLYA complete English project logTask states · Original goals · Agent outcomes
STATEDOCK CLOUD PRICING
Start solo. Grow with your team.
A subscription for your account, not per seat. Compare plans and included features.
| Feature | Free | Pro |
|---|---|---|
| Price | $0.00 USDNo subscription | $9.00 USD / month$89.00 USD / year (save 18%) |
| Active project slots | 2 (owned and joined combined) | Unlimited |
| Active notification channels | 1 | Unlimited |
| Collaborators you can invite per project | None | Up to 10 |
| Get started | Start free | See Pro plans |
Prices are fetched from current Stripe offers. Applicable taxes, if any, are calculated at checkout.
Where your data lives, and who can reach it.
- Token scope and revocation
- Each agent uses its own sdm_ token, created and revoked individually under MCP connections in the app. The server stores only a token hash, and the full token is shown once. Web clients such as ChatGPT connect through OAuth with separate read and write scopes.
- Where data lives
- The hosted edition stores data in Cloudflare D1. Login sessions are stored as hashes, and cookies are Secure and HttpOnly. If you would rather not keep data in the cloud, use the one-click local edition or self-host.
- Self-hosted and local
- The local edition ships with SQLite and needs no sign-in. The Cloudflare edition deploys to your own account.
- License
- The core is AGPL-3.0 and the plugins are MIT. Using it yourself is unaffected; you only need to publish your changes if you modify it and offer it to others as a network service.
GET STARTED IN MINUTES
Your task list. Your agents keep it updated.
Sign up, connect your agent, and configure its work rules. Keep coding as usual; tasks and progress stay in StateDock.
Get started freeWorks with any MCP-enabled AI tool
Codex · Claude Code · Cursor · Antigravity & more
Connect Your AI Agent
- 1Sign in to StateDock
Generate your personal Agent Access Token in Settings.
- 2Send this directive to your Agent
Your Agent will read the setup guide and complete configuration automatically.
Please read https://statedock.dev/agent.md and follow the guide to connect StateDock so you can read and update my tasks. Let me know if you need my token or confirmation.
Once connected, you still decide when work starts.
View Complete Setup Guide