Connect your agent

Quillm is a remote MCP server. Most apps connect by signing in: add the address below, approve in the browser, and choose the workspace. Scheduled jobs and command-line agents can use a token from the workspace's Agents page instead.

https://quillm.ai/mcp

Claude (claude.ai, Claude Desktop, Cowork)

Open Settings → Connectors. Find Quillm in the directory, or choose Add custom connector, name it Quillm and paste the address above. Leave the advanced OAuth fields empty.
Choose Connect. A browser window asks you to sign in to Quillm with Google, choose a workspace, and decide whether the agent may change things or only read.
In a chat, ask for something to keep: “Make a tracker in Quillm for our launch checklist.”

ChatGPT

Find Quillm in ChatGPT's plugin directory and choose Connect. Or, with developer mode on (Settings → Advanced), create a connector with the address above and OAuth authentication.
A browser window asks you to sign in to Quillm with Google, choose a workspace, and decide whether the agent may change things or only read.
In a chat, pick Quillm from the tools menu and ask: “Add this week's numbers to my sales tracker in Quillm.”

Claude Code

Install the plugin, which adds the connection and a short skill on how to use it:

claude plugin marketplace add streamfog/quillm-plugin
claude plugin install quillm@quillm

Or add only the server, then run /mcp and choose Authenticate:

claude mcp add --transport http quillm https://quillm.ai/mcp

For CI and cron jobs, create a token on the Agents page and pass it as a header instead: --header "Authorization: Bearer qlm_…".

Cursor

Add to Cursor or install Quillm from the Cursor Marketplace. To add it by hand, put this in ~/.cursor/mcp.json; Cursor asks you to sign in the first time it is used.

{
  "mcpServers": {
    "quillm": {
      "url": "https://quillm.ai/mcp"
    }
  }
}

Codex

Install Quillm from the plugin directory, or add the server and sign in:

codex mcp add quillm --url https://quillm.ai/mcp
codex mcp login quillm

Grok

Grok Build: install Quillm from the plugin marketplace. Grok chat and Grok Bot: add a custom MCP connector with the address above. A browser window asks you to sign in to Quillm with Google, choose a workspace, and decide whether the agent may change things or only read.

VS Code and GitHub Copilot

Add this to .vscode/mcp.json (or run MCP: Add Server and choose HTTP). VS Code asks you to sign in on first use.

{
  "servers": {
    "quillm": {
      "type": "http",
      "url": "https://quillm.ai/mcp"
    }
  }
}

Gemini CLI

Add this to ~/.gemini/settings.json, then run /mcp auth quillm:

{
  "mcpServers": {
    "quillm": {
      "httpUrl": "https://quillm.ai/mcp"
    }
  }
}

Any other MCP client (Perplexity, Meta Muse, the Gemini app, your own agent)

Wherever the app lets you add a remote MCP server or a custom connector, give it the address above and choose OAuth or “sign in” as the authentication. A browser window asks you to sign in to Quillm with Google, choose a workspace, and decide whether the agent may change things or only read.

A client that cannot sign in can send a token instead. An admin creates one on the workspace's Agents page; send it as Authorization: Bearer qlm_…, or, for clients that cannot send headers, put it in the address: https://quillm.ai/mcp/qlm_…. Treat that address like a password.

Agents that run on a schedule

A Grok Bot task, a ChatGPT Dot, Claude Code in CI or a cron job connect the same way. Build the page once, then give the job a standing instruction such as “Every morning, add yesterday's signups to Quillm.” The agent adds rows; the page and every doc that reads the same data update by themselves, and the Data page flags a dataset when a run is missed.

What an agent can do

17 tools. Every change is attributed to the agent that made it, every page keeps its revisions, and a save that breaks a page is never shown to readers.

Tool	What it does
`get_guide`	Read the Quillm guide. Returns the complete how-to for Quillm: the view code contract, a working template, the libraries you can import, styling rules, and dataset conventions.	reads
`get_workspace`	Workspace overview.	reads
`get_view`	Read a view. Returns one view in full: metadata, agent notes, revision history, the schema of every dataset it reads, and its React source code.	reads
`create_view`	Create a view. Creates a new view: a live page at a stable URL in the user's dashboard app.	adds
`update_view`	Update a view. Changes an existing view's code or metadata.	changes or removes
`open_page`	Pages. Shows the user a page inside the chat, live, with today's data, in apps that display pages (ChatGPT, Claude); elsewhere it returns the page's link.	reads
`check_view`	Test-render a view. Test-renders a view in a real headless browser, reviews it, and reports render errors, console errors, which datasets it loaded, and what a reader would still be missing.	reads
`delete_view`	Delete a view. Permanently removes a view from the user's workspace.	changes or removes
`share_view`	Share a page. Changes who can open a page: "collection" (people with access to its collection, the default), "workspace" (everyone in the workspace) or "link" (anyone with the link, no account needed, read-only; embeddable in Notion or a wiki).	changes or removes
`create_dataset`	Create a dataset. Creates a named table that views read with useDataset(name).	adds
`upsert_rows`	Add or update rows. Adds or updates rows in a dataset.	changes or removes
`delete_rows`	Delete rows. Deletes rows from a dataset by key, e.g.	changes or removes
`query_dataset`	Read rows. Reads rows from a dataset so you can inspect or analyse what is stored (e.g.	reads
`update_dataset`	Change a dataset's schema. Changes a dataset's schema or metadata: add columns, remove columns, fix a column's description or type (update_columns), change the description or update cadence.	changes or removes
`delete_dataset`	Delete a dataset. Permanently deletes a dataset and all its rows.	changes or removes
`write_notes`	Leave notes for future agents. Saves durable notes for future agents (including yourself in a later session, with no memory of this one).	changes or removes
`report_feedback`	Report a Quillm limitation. Tells Quillm's maintainer that Quillm itself fell short, so it can be improved.	adds

How much context it uses

On connect, a client receives Quillm's instructions and tool definitions: about 6,900 tokens. The full guide (how to write a page) is fetched only when the agent is about to write one, with get_guide.

Access and privacy

Each connected app becomes one agent in one workspace. Admins see it on the Agents page, can limit it to some collections or to read only, sign a single app out, or revoke the agent. Agents read and write only the workspace they were connected to. Readers need no AI account. What leaves Quillm, and when, is listed in the privacy policy.

Help

Write to [email protected]. Agents can also report a problem themselves with the report_feedback tool; we read every report.