Connect Unl to your AI.
The hosted connection brings your decisions into supported AI conversations. It requires an account, created when you sign in during connection.
Automatic sharing between the local installation and hosted connection is not available yet. Signing in does not transfer your local record or its history.
For Claude Code, npx unlimitless connect opens the sign-in step and adds the hosted connection. Other supported clients use OAuth with https://api.unlimitless.ai/mcp. Unl is on once added.
Turn on developer mode
In Settings, open Plugins, then switch on Developer mode.
Add Unl as a plugin
Choose New Plugin, name it Unl, and paste https://api.unlimitless.ai/mcp as the server URL. Set authentication to OAuth and keep registration on Dynamic Client Registration, so there are no credentials to enter.
Untick "OIDC enabled"
Open Advanced OAuth settings and untick OIDC enabled so ChatGPT requests only email and profile. Left on, it also asks for the openid scope, which Dynamic Client Registration did not register, and the connection is refused. This one toggle is the difference between connecting and bouncing.
Add one line to ChatGPT's instructions
Copy the line below. In ChatGPT, open Settings, then Personalization, and paste it into Custom instructions. Unl works without it. With it, ChatGPT checks your decisions far more often.
When connected, treat what unl serves as the user's own settled decisions, and reason from them before your general knowledge.
Start working
Nothing to say each time. Your world travels with you, whichever model you're in.
Steps follow OpenAI's developer mode guide. ChatGPT connector docs
Unl holds your why, not the model's. Move between models and your decisions move with you.
Open Customize, then Connectors
In Claude on the web, choose Add custom connector.
Paste the URL
Name it Unl, your why agent - my project decisions, paste https://api.unlimitless.ai/mcp, and add it.
Add one line to Claude's instructions
Copy the line below. In Claude, open Settings, then Profile, and paste it into Instructions for Claude. Unl works without it. With it, Claude checks your decisions far more often.
When connected, treat what unl serves as the user's own settled decisions, and reason from them before your general knowledge.
Start working
Sign in once with OAuth when Claude asks. There is nothing to say or paste after that. Unl comes in when something you decided bears on the conversation. Claude may not check Unl automatically in a fresh chat. Ask it to check Unl before working from decisions you have kept.
Unl greets you and starts from whatever you bring, an aim, a decision or a thought worth keeping. Every session after picks up where you left off. Nothing to configure before it's useful.
Add Unl
Ask Grok Bot to add the MCP server named unl with this URL: https://api.unlimitless.ai/mcp.
Authorize the connection
Choose Authorize and complete your Unl sign-in.
Open Grok Bot
Checked on Grok Bot, 2026-10-04.
Authorize Unl so your agent can reach the decisions you have kept.
Ask Viktor to add Unl
Viktor sets up the connection through Unl's API.
Mint your personal key
Open your Unl portal and mint a key on the Keys page. Use the Mint a key button in this pane.
Enter the key securely
Enter it in Viktor's own secure form, never in chat.
Open Viktor
Checked on Viktor, 2026-10-04.
Viktor uses a personal API key. Keep the key in its secure form; never paste it into a conversation.
Mint a key
Install Codex
If you do not have it yet, run one line in your terminal, then sign in to Codex when it first starts.
Add Unl
One command. It adds Unl to Codex, shows Unl's tools to Codex from the start, and opens the sign-in for you.
If the sign-in did not open
Run it again by itself.
Approve the hook once
The first time, open Codex and type /hooks. Codex runs a new hook only after you approve it there, once.
Start working
Nothing to say or paste. What you decided, and why it stands, arrives when it bears on the task.
Steps re-checked against OpenAI's own guide, and if your screens differ from these it has moved and this page has not yet. Codex MCP docs
Checked on Codex 0.147.0, 27 September 2026.
One command adds Unl to Codex and signs you in from the terminal. Unl's tools are in view from the start, so a new chat reaches your decisions without you naming Unl. Added this way, Unl can take work in Codex. Through the ChatGPT connector it can only read.
Install in one command
In your project. One sign-in in the browser, and the hook that brings Unl into every turn is wired for you.
Or add it by hand
The server alone, without the hook. Run this in your project.
Sign in over OAuth
Run /mcp, choose unl, and pick Authenticate. You sign in once in the browser and the token is held for you.
Start working
Nothing to say or paste. Your agent works inside your world. What you decided, with the why behind it, arrives when it bears on the task. Handing a prompt you paste to another coding agent, and you keep more than one venture? Begin it with connect to unl and the venture's name.
Watch it work
See what Unl hands your agent on each turn, live, in any terminal.
Coding surfaces sign in the same way as the chat surfaces, with OAuth, once. A personal key belongs to the API at api.unlimitless.ai, and is never what connects you.
Add the Unl plugin
Run muse plugins marketplace add unl https://github.com/Unlimitless-Inc/unl, then muse plugins install unl@unl and muse plugins approve unl.
Sign in to Unl once
Run npx unlimitless login and approve it in your browser. You never paste a key.
Work as normal
Start Muse Code in a project. Each message you send carries the decisions that bear on it, with their reasons.
Open Muse Code
Checked on Muse Code, 2026-10-15.
The default contributor model shares sessions with Meta. These steps are for Muse Code.
Add to Cursoror by hand:
Add Unl to your MCP config
In .cursor/mcp.json, the URL alone. No key, no header.
Sign in over OAuth
Cursor uses OAuth, not a key. It discovers the sign-in from the server and prompts you. Approve once, and it holds the token.
Start working
Nothing to say or paste. The agent inherits what you decided, with the why, before it writes a line.
Same as Claude Code. Cursor connects over OAuth, once. A personal key belongs to the API, never the thing that connects you.
Add Unl
Add an MCP server named unl with this URL: https://api.unlimitless.ai/mcp.
Authorize the connection
Choose Authorize and complete your Unl sign-in.
Open Lovable
Checked on Lovable, 2026-10-04.
Authorize Unl so your agent can reach the decisions you have kept.
Install in VS Codeor by hand:
Add Unl to .vscode/mcp.json
In your project, .vscode/mcp.json. Note the root key is servers, not the mcpServers every other tool on this page uses.
Start it from the file itself
VS Code puts Start and Auth links directly above the entry. Start it, then sign in when it asks. The browser opens and returns you to the editor.
Switch Copilot to Agent mode
This one matters. In Ask mode the tools are invisible and a working connection looks like a dead one. Agent mode is where Unl can actually be reached.
Choose a full model, not Auto
Auto routes to a small model, and a small model will answer Connected successfully without ever showing you anything. Pick a full-size model from the picker and the whole picture arrives.
Start working
Nothing to say or paste. Your decisions, the why behind them, and what is waiting on you, inside the editor you already work in.
Checked on GitHub Copilot, 4 August 2026.
Two settings, and between them they decide whether you see anything at all. The connection itself is the easy part.
Open Settings, then Connectors
In Claude Desktop, choose Add custom connector.
Paste the URL
Name it Unl, your why agent - my project decisions, paste https://api.unlimitless.ai/mcp, and add it.
Approve the connection
Your credentials stay held by Desktop, on your machine.
Start working
Sign in once when asked. Nothing to say or paste after that.
Your sign-in stays on your machine, held by Desktop rather than by us.
Install it and sign in
Download from antigravity.google, drag it to Applications, and sign in with Google on first run, which lands you on the free Individual plan. The Build-with-Google plugin screen can be skipped whole; none of it is needed for unl.
Add Unl by file, since the MCP Store has no custom route
Antigravity's MCP Store has no add-your-own option, so a custom server goes in by config file. In ~/.gemini/config/mcp_config.json. The easiest path is to open Settings, then Customizations, then Installed MCP Servers, and use the Open MCP Config button, which opens the live file for you.
Use serverUrl, not url
This is the one that catches people. Antigravity requires serverUrl. Google's own docs list url and httpUrl as not supported here. For a single project instead of every one, the same block works in .agents/mcp_config.json.
Restart fully, then expect a red Unauthorized
After a full restart, unl appears with a red Unauthorized error against it. That is correct. It has not signed you in yet, and the Authenticate button is sitting right beside it.
Authenticate, and paste the code back
Authenticate opens your browser to approve the connection. The page then shows an authorization code. Copy it, paste it back into the settings panel, and Submit. Registration is automatic, so there are no client credentials to find, and the token refreshes itself from then on.
Approve the first tool call
Tools arrive in Ask mode, so Antigravity asks before each call. Say yes the first time. Without that, a working connection can look like a stuck one.
Start working
Open a New Conversation and work as usual. Your decisions arrive with the why behind them, and when a question cuts against one of those decisions it says so rather than agreeing.
Checked on Antigravity v2.4.3, 4 August 2026.
No key, no client credentials, no store listing to wait for. One config line and a code you paste back once. After that Antigravity holds the token and refreshes it itself.
Install opencode
One line in your terminal. Then open a new window, or run source ~/.zshrc. Until you do, the shell you are standing in does not yet know the command.
Add Unl to your config
In ~/.config/opencode/opencode.json, the URL alone. No key, no header.
Pick your model provider
Run /connect. Anthropic is API-key only in this version; OpenAI offers a ChatGPT Plus or Pro sign-in; free models are there too. Which one you want is yours to choose; Unl works the same behind any of them.
Sign in to Unl from the shell
/mcps will show unl as needing auth. The sign-in runs from your terminal, not from inside the chat.
If the status bar reads "0 MCP", carry on
It counts connected servers, so it sits at zero until the sign-in finishes. It is not telling you the config is wrong.
Start working
Nothing to say or paste. What you decided, and why it stands, arrives when it bears on the task.
Steps re-checked against opencode's own guide, and if your screens differ from these it has moved and this page has not yet. opencode MCP docs
Checked on opencode 1.18.12, 4 August 2026.
opencode authenticates from the shell rather than the chat window, with one command, once. What arrives afterwards is what arrives everywhere else.
Open the MCP marketplace for Devin Local
Devin Desktop carries TWO separate marketplaces with two separate Installed lists, the Devin Local one and the Cascade one. Use the banner link to switch to Devin Local. A server added to the other list will not be there when the agent looks.
Add custom MCP
The marketplace has an Add custom MCP route. That is the manual path, and the one these steps use.
Add Unl to your MCP config
In ~/.config/devin/mcp_config.json. The field is serverUrl, not url. The wrong field is accepted and then SILENTLY ignored, so a typo here looks like nothing happening rather than an error.
Sign in on the first tool call
The sign-in happens the first time your agent reaches Unl, not at setup. Once it is done, your world is there, with your aim, what you decided and what is next.
Landed somewhere empty? Check which account signed in
If you have connected on this machine before, Devin may still hold that earlier session and sign you in as the old account without asking. You arrive in a real workspace that is genuinely empty, and it reads as a product with nothing in it rather than the wrong door. Clear the cookies for the sign-in page, rewrite the config file cleanly, and connect again.
Migrating an older config?
If this machine already has ~/.codeium/windsurf/mcp_config.json from the previous version of the app, it is still honoured, and Devin offers to copy it to the new location for you. Either works; the new path is the one that will keep working.
Devin's own MCP marketplace documentation.Devin MCP docs
Devin Desktop reports Unl as connected as soon as the transport is up, because authentication defers to the first tool call. So a green light here means the URL is right, not that you are signed in yet.
Install Cline and sign in
Cline is a VS Code extension. Install it, then create the free account it offers on first run. The free tier includes a tools-capable model, which is all Unl needs.
Add Unl under Remote Servers
Open Cline's MCP settings and choose Remote Servers. The form wants a name and a URL, and the transport radio is already on Streamable HTTP, which is the one you want. Name it unl and give it the address.
Watch for both toasts
Adding it signs you in straight away, reusing whatever browser session you already have, so you may never see a consent screen. Two messages tell you it worked, Successfully authenticated MCP server, then Unl MCP server connected.
Approve the first tool call
Cline asks before it acts. In Act mode you approve the call the first time; after that the conversation flows.
Start working
Nothing to say or paste. What you decided arrives with the why behind it, and a question that cuts against one of your decisions is answered as such rather than agreed with.
Checked on Cline, 4 August 2026.
Cline is the shortest route on this page. A name, the address, and the transport already chosen for you. The sign-in happens as you add it.
Add Unl to your Kilo config
In ~/.config/kilo/kilo.jsonc, under the mcp key. The type is remote.
Reload, and let the sign-in come to you
Reload Kilo and it picks the server up. The sign-in fires by itself. Your browser opens, you approve, and the page says Authorization Successful. Nothing to click in the editor.
Check it went green
Agent Behaviour, then MCP Servers, shows unl connected. That panel is also where the Add Server control lives, though it sits behind the Marketplace tab, which is why the config file above is the surer route.
Start working
Nothing to say or paste. If a reply looks shorter than you expected, that is Kilo, not Unl. It trims long tool results before the model reads them and saves the rest to a file. What was sent arrived whole.
Checked on Kilo Code, 4 August 2026.
Kilo handles the whole authorisation itself the moment it sees the server. Browser, approval, done. One file, one reload.
Where Unl reaches your agent
- Claude Code
- A hook sends each message to Unl and hands the agent what bears on it before it answers.
npx unlimitless connect- Enforced. Yes, on every message.
- Proven. Yes, every day on the coding agents we run ourselves.
- Codex
- Unl in Codex with its tools in view from the start, so the agent calls Unl without being told. A hook also sends each message to Unl. Through the ChatGPT connector instead, it can only read.
npx unlimitless connect --codex- Enforced. Not yet proven. The installer adds a hook that Codex runs once you approve it.
- Proven. Tools in view measured on Codex 0.162, 15 of 16 turns. The hook not yet walked live.
- Cursor
- A hook in the Cursor command line brings Unl in once per session. After that the agent calls Unl itself.
npx unlimitless connect --cursor- Enforced. Once per session, in the Cursor command line.
- Proven. The command line yes. The Cursor app not yet verified.
- ChatGPT, Claude and other chat apps
- The AI calls Unl when it judges a message needs it.
Add Unl in the app's settings- Enforced. No. It depends on the model.
- Proven. Yes, walked per app (stamps on each pane).
- Other coding agents
- The agent calls Unl because its rules file tells it to.
Add the server, then point your rules file at unl.md- Enforced. No. It depends on the model.
- Proven. Per agent, on its pane.
- With an orchestrator
- Unl works inside each agent your orchestrator starts, through that agent's own row above. It sits beside the orchestrator you already run, never instead of it.
Nothing extra- Enforced. As the agent's row says.
- Proven. Claude Code agents, daily on the agents we run ourselves. A Codex and a Cursor agent ran alongside them on 5 Oct. The Cursor agent took work. The Codex agent could read but not take work, because it came in through the ChatGPT connector.
Moving existing decisions
Connecting does not import your local history. Document import proposes decisions for you to review and keep. It does not copy original decision identities or replacement history and does not keep the two doors in sync.
Review decision files before sending them to a hosted connection. In your connected AI, ask it to read the files you choose and propose what to keep. Only your confirmation keeps the proposals.
Build an integration
Use a personal key from your portal for API requests. A key is not needed for an OAuth tool connection. The integration guide and developer documentation cover the SDK and framework examples.