MCP server
Connect Claude.ai, ChatGPT, Claude Code, Cursor and other AI tools to your intoCHAT account through its remote MCP server: the endpoint, signing in or API keys, setup for each client, the tools, and the limits.
intoCHAT has a remote MCP server. MCP (Model Context Protocol) is the standard way AI tools such as Claude.ai, ChatGPT, Claude Code and Cursor connect to other services. Once you connect yours, you can ask it in plain words to list your agents, read their conversations and leads, add text snippets and Q&A pairs, delete sources or send an agent a test message, and it calls intoCHAT for you.
There are two ways to connect:
- Sign in (OAuth). For connectors in Claude.ai and ChatGPT: you paste the server's address, sign in to intoCHAT and choose what the app may do. No key to copy. See Add intoCHAT to Claude.ai and Add intoCHAT to ChatGPT.
- An API key. For tools that let you set a header, such as Claude Code and Cursor. It uses the same API keys, scopes and rate limit as the REST API.
Either way, the tools do exactly what the matching REST endpoints do.
Endpoint
https://www.intochat.ai/api/mcp
The server uses the Streamable HTTP transport. Every request needs either your API key or an access token from signing in, in the Authorization header:
Authorization: Bearer ic_live_YOUR_KEY
Apps that sign in get their access token themselves; you don't see it. Use the address with www, as above.
Before you start
The MCP server is included on Starter, Pro and Enterprise, like the REST API. On Free you can't create API keys or connect apps, and an account that moves back to Free gets an error on every request until it's on Starter or higher again.
- To connect by signing in, you need an intoCHAT login with access to the account. Anyone on the team can connect an app; it can never do more than they can. See What a connected app can do.
- To connect with an API key, you need a key. Only the account owner and admins can create one.
Add intoCHAT to Claude.ai
-
In Claude.ai, open Settings, then Connectors, and click Add custom connector. Depending on your Claude plan, an organization owner may have to add it for everyone instead.
-
Give it a name, for example
intoCHAT, and paste the server's address as the URL:https://www.intochat.ai/api/mcp -
Click Add, then Connect. A new window opens on intoCHAT.
-
Sign in to intoCHAT if you aren't signed in yet. You come back to the consent screen afterwards.
-
On the consent screen, check the app's name and the address it sends you back to, then choose:
- Account, if you belong to more than one. An account on Free can't be connected.
- Access: Read only, or Read and write.
- Agents: All agents, or Only the agents I choose.
-
Click Allow. You're sent back to Claude.ai and the connector shows as connected.
Then ask Claude, for example: "List my intoCHAT agents." Custom connectors you add in Claude.ai also work in the Claude desktop app.
Add intoCHAT to ChatGPT
- In ChatGPT, open Settings, then Apps & Connectors. Custom connectors need Developer mode, under Advanced settings. Depending on your ChatGPT plan, they may not be available, or a workspace admin may have to allow them.
- Click Create and fill in:
- Name: for example
intoCHAT. - MCP server URL:
https://www.intochat.ai/api/mcp - Authentication: OAuth.
- Name: for example
- Click Create. ChatGPT opens intoCHAT's sign-in.
- Sign in to intoCHAT, choose the account, access and agents on the consent screen as in step 5 above, and click Allow.
Then pick the intoCHAT connector in a chat and ask, for example: "Which of my intoCHAT agents had the most conversations this week?"
What a connected app can do
On the consent screen you choose what the app may do. It never gets more than that, and never more than you can do in the account yourself.
- Read only: the tools that read:
list_agents,get_agent,list_conversations,get_conversation,list_inbox_conversations,list_leads,list_contactsandlist_sources. - Read and write: also
chat_with_agent(each reply counts as one message on your plan with the default model, more with a larger model),add_text_source,add_qa_pair,delete_sourceandupsert_contact. Not offered if the app asked for read access only. - Your role applies. A Viewer who allows read and write still can't add or delete sources or update contacts: those tools answer with an error naming your role. If your role changes, the app's access changes with it on its next request.
- Your agents apply. If you work on selected agents only, the app reaches only those, even with All agents. With Only the agents I choose, it reaches the ones you ticked, as long as you still have access to them.
- If you're removed from the account or leave it, the apps you connected to it stop working.
The app keeps working without asking you again: its access renews itself while it's in use. If it isn't used for 30 days, connect it again.
Disconnect an app
- In intoCHAT, go to Settings, Account.
- Under Connected apps, find the app. Each one shows its access, its agents, the account it uses if it isn't your own, and when it was last used.
- Click Revoke, then Revoke access.
The app loses access immediately. Connected apps lists the apps you connected yourself, in any account you belong to. Connecting and disconnecting apps appear in the account's audit log as "App connected" and "App disconnected".
Connect with an API key
The rest of this section is for tools that send a header instead of signing in.
Create an API key
- In the sidebar, under Settings, click API keys.
- Under Create a key, give the key a Name, for example
Claude Code. - Under Access, choose what the AI tool may do. Every scope, including ones added later allows every tool. To keep it read-only, pick Only the scopes I choose and tick only the
:readscopes. See Tools for the scope each tool needs. - Click Create key, then Copy, and click I've saved it. The key is shown only once.
A key works for every agent in the account. Treat it like a password: keep it out of files you commit or share, and revoke it on the API keys page if someone else may have it. See Create an API key in the REST API docs for more.
Connect Claude Code
Run this in a terminal, with your key in place of ic_live_YOUR_KEY:
claude mcp add --transport http intochat https://www.intochat.ai/api/mcp \
--header "Authorization: Bearer ic_live_YOUR_KEY"
This adds the server for the current project, on your computer only. To use it in every project, add --scope user before --transport.
To share the setup with your team in the project's .mcp.json without the key itself, read the key from an environment variable:
{
"mcpServers": {
"intochat": {
"type": "http",
"url": "https://www.intochat.ai/api/mcp",
"headers": {
"Authorization": "Bearer ${INTOCHAT_API_KEY}"
}
}
}
}
Then type /mcp in Claude Code to check that intochat is connected, and ask, for example: "List my intoCHAT agents."
Connect Cursor
In Cursor's settings, open the MCP section and add a server, or edit ~/.cursor/mcp.json (all projects) or .cursor/mcp.json in a project:
{
"mcpServers": {
"intochat": {
"url": "https://www.intochat.ai/api/mcp",
"headers": {
"Authorization": "Bearer ic_live_YOUR_KEY"
}
}
}
}
The intoCHAT tools then appear in the list of MCP tools, and Cursor's agent can use them.
Other clients
Any MCP client that supports the Streamable HTTP transport and lets you add a header can connect. Give it the endpoint and the Authorization header. Clients that support MCP's OAuth sign-in can connect without a key, as in Claude.ai.
Claude desktop app. Add intoCHAT under Connectors, as in Claude.ai. To use an API key instead, add the server to claude_desktop_config.json through the open-source mcp-remote bridge, which needs Node.js on your computer:
{
"mcpServers": {
"intochat": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://www.intochat.ai/api/mcp", "--header", "Authorization:${INTOCHAT_AUTH}"],
"env": {
"INTOCHAT_AUTH": "Bearer ic_live_YOUR_KEY"
}
}
}
}
MCP Inspector. Run npx @modelcontextprotocol/inspector, choose Streamable HTTP, enter the endpoint and add the Authorization header. Connect through the Inspector's proxy: the server refuses requests made directly from a web page on another site.
Testing with curl. Every request is one JSON-RPC message in a POST:
curl https://www.intochat.ai/api/mcp \
-H "Authorization: Bearer ic_live_YOUR_KEY" \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-H "MCP-Protocol-Version: 2025-11-25" \
-d '{"jsonrpc": "2.0", "id": 1, "method": "tools/list"}'
The server speaks MCP versions 2026-07-28 and 2025-03-26 to 2025-11-25, answers every request with JSON, and keeps no session between requests.
Tools
| Tool | Scope | What it does |
|---|---|---|
list_agents | agents:read | Lists the account's agents with their IDs, training status and counts. |
get_agent | agents:read | Gets one agent. |
chat_with_agent | chat | Sends a message to an agent and returns its reply. Each reply counts as one message on your plan, or more with a larger model. |
list_conversations | conversations:read | Lists an agent's conversations, newest first, without transcripts. |
get_conversation | conversations:read | Gets one conversation with its transcript. |
list_leads | leads:read | Lists the leads an agent collected, newest first. |
list_contacts | leads:read | Lists an agent's contacts, newest first, or looks one up by email or external_id. |
upsert_contact | contacts:write | Creates or updates a contact by external_id, else email, with phone, name and attributes (an object; null removes an attribute). |
list_sources | sources:read | Lists an agent's knowledge sources, its training progress and the characters it uses. |
add_text_source | sources:write | Adds a text snippet with a title and starts training. |
add_qa_pair | sources:write | Adds a question and answer and starts training. If the agent already has the same question, its answer is replaced. |
delete_source | sources:write | Deletes a source and what the agent learned from it. This can't be undone. |
list_inbox_conversations | conversations:read | Lists an agent's conversations in the Inbox: open ones by default, or those with the status you pass (pending, snoozed or solved), with their inbox fields. |
Each tool returns the same data as its REST endpoint, as JSON. list_conversations, list_inbox_conversations, list_leads and list_contacts take an optional limit (1 to 100), cursor and since, as described under Pagination.
The tools are marked as read-only or not, and delete_source, add_qa_pair and upsert_contact as able to change what's already there, so your AI tool can ask you before it uses them. Many AI tools ask for your confirmation before they run any tool.
If the key doesn't have the scope a tool needs, the tool answers with an error that names the scope, and nothing happens. For a connected app, the error says whether the app was connected read-only or your role doesn't allow the tool. See What a connected app can do. An agent, conversation or source in another account is "not found", the same as one that doesn't exist.
There are no tools to create agents or change their settings or instructions. Do that in the dashboard.
Chatting with an agent
chat_with_agent sends a real message, answered by the same code as the chat widget, from the agent's knowledge, instructions and actions. Use it to test how an agent answers.
- Each reply counts as one message on your plan (more with a larger model) and toward the agent's monthly message cap, if you set one. See What counts as a message.
- Leave out
conversationIdto start a new conversation. To continue one, pass theconversationIdfrom the earlier reply. - The conversations appear on the agent's Conversations tab like any other.
- A private agent can't be reached this way. See Private agents.
Ask your AI tool not to send many messages in a row unless you mean it to: each one uses your plan's messages.
Limits
- Plan: Starter, Pro or Enterprise. On Free, requests are refused.
- Rate limit: 60 requests a minute per key, shared with the REST API, or per connected app. Connecting takes a few requests, and so does every tool call. Over the limit, requests are refused for the rest of the minute. A separate key for your AI tool keeps it from using your server's requests.
- Messages: each
chat_with_agentreply counts as one message on your plan, or more with a larger model. A reply must finish within 55 seconds. - Sign-in: an API key, or OAuth sign-in with dynamic client registration and PKCE, for clients such as Claude.ai and ChatGPT. Apps that sign in must send you back to an
httpsaddress, or tolocalhoston your own computer. - Sources: text snippets and Q&A pairs only. Add website pages, files and Notion pages in the dashboard.
- No live updates: the server answers each request and doesn't push notifications.
If your client can't connect, check the error it shows: an invalid or revoked key, or a connection that was revoked or expired, is refused with 401, an account on Free with 403, and too many requests with 429. A connected app that gets 401 usually offers to connect again.
Next steps
- Call intoCHAT from your own code: REST API.
- Get leads and hand-offs pushed to your server as they happen: Webhooks.
- Write good text snippets and Q&A pairs: Text and Q&A.