How to integrate Ollama MCP with Kimi Code

Trusted by
AWS
Glean
Zoom
Airtable
GET STARTED FOR FREEGET A DEMO

30 min · no commitment · see it on your stack

Ollama logo
Kimi Code logo
divider

How to integrate Ollama MCP with Kimi Code

Kimi Code is Moonshot AI's open-source coding agent, powered by Kimi K2.6. It runs in your terminal, reads and edits code, executes shell commands, and plans multi-step tasks, with native MCP support for extending it to outside tools.

In this guide, I will explain the easiest and most secure way to connect your Ollama account to Kimi Code via Composio Connect, so it can run a custom prompt with llama2, get model info for all installed models, generate a text completion using mistral, and more without ever putting your account credentials at risk.

Also integrate Ollama with

Why use Composio?

Composio provides:

  • Access to 1,000+ managed apps from a single MCP endpoint. This makes it convenient for agents to run cross-app workflows.
  • Managed OAuth. You do not have to worry about authentication and authorization flows for every app.
  • Programmatic tool calling. Allows LLMs to write code in a remote workbench to handle complex tool chaining. This reduces back-and-forth for frequent tool calls.
  • Large tool response handling outside the LLM context. This minimizes context bloat from large tool responses.
  • Dynamic just-in-time access to thousands of tools across hundreds of apps. Composio loads the tools your agent needs, so LLMs are not overwhelmed by tools they do not need.

Connect Ollama to Kimi Code

Kimi Code is a TypeScript agent distributed through npm. It acts as an MCP client and reads server definitions from an mcp.json file, and it can also add and authenticate servers conversationally through /mcp-config. Composio is a remote HTTP server that authenticates with OAuth, so no API key is stored anywhere.

1. Install Kimi Code

The quickest way is the official install script, which requires no pre-installed Node.js and places the kimi executable on your PATH.

bash
# macOS or Linux
curl -fsSL https://code.kimi.com/kimi-code/install.sh | bash

# Windows PowerShell
irm https://code.kimi.com/kimi-code/install.ps1 | iex

# Confirm the installation
kimi --version

2. Log in

Start Kimi Code in your project directory, then sign in from the interactive UI:

bash
kimi

Run /login and choose Kimi Code OAuth using the device-code flow, or use a Moonshot API key.

3. Add Composio with /mcp-config

In current versions of Kimi Code, MCP servers are managed inside the app, not with a shell subcommand. From the interactive UI, run:

bash
/mcp-config
Kimi Code MCP config flow for adding the Composio MCP server

Tell it the server name and URL in plain language. For example:

Server name is Composio, and here is the server URL: https://connect.composio.dev/mcp

Kimi Code asks whether to add it globally, at ~/.kimi-code/mcp.json, or project-local for the current checkout, then writes the entry for you:

bash
{
  "mcpServers": {
    "Composio": {
      "url": "https://connect.composio.dev/mcp"
    }
  }
}

There is no transport field to set. Kimi Code infers HTTP from the url.

4. Restart the session

The new server is picked up on a fresh session, not the current one. Start a new session:

bash
/new

On the new session, Kimi Code detects that the server needs authorization and prompts you to run:

bash
/mcp-config login Composio

5. Authorize with OAuth

Run the command Kimi suggests:

bash
/mcp-config login composio

Kimi Code opens Composio's authorization page or surfaces a URL. Approve access, then return to the session. You should see confirmation that the Composio MCP server is connected.

Composio authorization page for Kimi Code MCP setup

Check the connection status any time with /mcp. Composio should appear as connected with its tools listed.

Kimi Code showing Composio connected after OAuth authorization

Connect your Ollama account

Back in a Kimi Code session, ask the agent to connect to Ollama or give it any Ollama-related task.

For example, ask it to:

  • "Run a custom prompt with llama2"
  • "Get model info for all installed models"
  • "Generate a text completion using mistral"

It will prompt you to authenticate and authorize access to Ollama.

That is it. Composio tools are now available in Kimi Code, and your Ollama account is ready to use.

Supported Tools & Triggers

Tools
Chat with Ollama modelTool to send a chat message with conversation history to Ollama.
Generate Text with OllamaTool to generate text responses from Ollama models with optional raw mode.
List ModelsTool to list all available Ollama models and their details.
OpenAI-Compatible Chat CompletionTool to create OpenAI-compatible chat completions using Ollama models.
OpenAI-Compatible Text CompletionTool to create OpenAI-compatible text completions using Ollama models.
List Models (OpenAI Compatible)Tool to list available models using OpenAI-compatible API format.
Show Model InformationTool to show comprehensive information about an Ollama model.
Get Ollama VersionTool to get the version of Ollama running locally.

Conclusion

You have successfully connected Ollama to Kimi Code using Composio Connect. Your agent can now manage Ollama from the terminal with natural language, without exposing credentials in prompts or local scripts.

Since the same Composio endpoint exposes 1,000+ apps, you can add Slack, Calendar, Linear, and more to the same server and chain them into cross-app workflows.

How to build Ollama MCP Agent with another framework

ChatGPT logo

ChatGPT

Use Ollama MCP with ChatGPT

Antigravity logo

Antigravity

Use Ollama MCP with Antigravity

OpenAI Agents SDK logo

OpenAI Agents SDK

Use Ollama MCP with OpenAI Agents SDK

Claude Agent SDK logo

Claude Agent SDK

Use Ollama MCP with Claude Agent SDK

Claude Code logo

Claude Code

Use Ollama MCP with Claude Code

Claude Cowork logo

Claude Cowork

Use Ollama MCP with Claude Cowork

Codex logo

Codex

Use Ollama MCP with Codex

Cursor logo

Cursor

Use Ollama MCP with Cursor

VS Code logo

VS Code

Use Ollama MCP with VS Code

OpenCode logo

OpenCode

Use Ollama MCP with OpenCode

OpenClaw logo

OpenClaw

Use Ollama MCP with OpenClaw

Hermes logo

Hermes

Use Ollama MCP with Hermes

CLI logo

CLI

Use Ollama MCP with CLI

Google ADK logo

Google ADK

Use Ollama MCP with Google ADK

LangChain logo

LangChain

Use Ollama MCP with LangChain

Vercel AI SDK logo

Vercel AI SDK

Use Ollama MCP with Vercel AI SDK

Mastra AI logo

Mastra AI

Use Ollama MCP with Mastra AI

LlamaIndex logo

LlamaIndex

Use Ollama MCP with LlamaIndex

CrewAI logo

CrewAI

Use Ollama MCP with CrewAI

FAQ

What are the differences in Tool Router MCP and Ollama MCP?

With a standalone Ollama MCP server, the agents and LLMs can only access a fixed set of Ollama tools tied to that server. However, with the Composio Tool Router, agents can dynamically load tools from Ollama and many other apps based on the task at hand, all through a single MCP endpoint.

Can I use Tool Router MCP with Kimi Code?

Yes, you can. Kimi Code fully supports MCP integration. You get structured tool calling, message history handling, and model orchestration while Tool Router takes care of discovering and serving the right Ollama tools.

Can I manage the permissions and scopes for Ollama while using Tool Router?

Yes, absolutely. You can configure which Ollama scopes and actions are allowed when connecting your account to Composio. You can also bring your own OAuth credentials or API configuration so you keep full control over what the agent can do.

How safe is my data with Composio Tool Router?

All sensitive data such as tokens, keys, and configuration is fully encrypted at rest and in transit. Composio is SOC 2 Type 2 compliant and follows strict security practices so your Ollama data and credentials are handled as safely as possible.

Used by agents from

Context
Letta
glean
HubSpot
Agent.ai
Altera
DataStax
Entelligence
Rolai
Context
Letta
glean
HubSpot
Agent.ai
Altera
DataStax
Entelligence
Rolai
Context
Letta
glean
HubSpot
Agent.ai
Altera
DataStax
Entelligence
Rolai

Never worry about agent reliability

We handle tool reliability, observability, and security so you never have to second-guess an agent action.

Get started for freeGet a demo↗