Ask Ollama MCP Server targets developers who want a model second opinion in a fully local setup, wiring local Ollama into MCP clients over stdio with no API keys. Its strengths are privacy, zero cost, and offline operation, plus sessionId-based conversation replay; its limits are that local models are weaker than frontier cloud models and Ollama with the target model must already be installed. Best for privacy-sensitive users with local compute, not for environments lacking Ollama or needing multi-provider parallel orchestration.
Strongest · Documentation 12/20 Weakest · Reliability 8/20
Static review · not runReviewed 2026-09-25
Read the FMRS scoring method →Runtime：Node.js 20+
OLLAMA_HOST
Address of the Ollama server, default http://localhost:11434; no change needed after installing Ollama locally.
GMCPT_TIMEOUT_MS
Timeout for Ollama execution in milliseconds, default 300000 (5 minutes).
GMCPT_LOG_LEVEL
Log verbosity: debug, info, warn, or error; default warn.
{
"mcpServers": {
"ollama": {
"command": "npx",
"args": [
"-y",
"@ask-llm/ollama-mcp"
],
"env": {
"OLLAMA_HOST": "http://localhost:11434",
"GMCPT_TIMEOUT_MS": "300000",
"GMCPT_LOG_LEVEL": "warn"
}
}
}
}
Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.
{
"servers": {
"ollama": {
"command": "npx",
"args": [
"-y",
"@ask-llm/ollama-mcp"
],
"env": {
"OLLAMA_HOST": "http://localhost:11434",
"GMCPT_TIMEOUT_MS": "300000",
"GMCPT_LOG_LEVEL": "warn"
}
}
}
}
Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).
claude mcp add ollama -e OLLAMA_HOST=http://localhost:11434 -e GMCPT_TIMEOUT_MS=300000 -e GMCPT_LOG_LEVEL=warn -- npx -y @ask-llm/ollama-mcp
Run it in a terminal; replace any <…> placeholders with your own values first.
The client's tool list should show ask-ollama, get-usage-stats, and ping; sending any prompt through ask-ollama and getting a reply from your local Ollama model confirms the connection works.
Once connected, you can ask your AI assistant things like:
No matching tools
Listed from the project's documentation, not tested by this site.
Ask Ollama MCP Server is the Ollama-only MCP server from the Ask LLM project (npm package @ask-llm/ollama-mcp), speaking stdio transport to MCP clients. It connects locally running Ollama models to clients such as Claude Code, Claude Desktop, Cursor, and Codex CLI so the primary model can call a local model for code review, plan debate, or a second opinion. Because inference happens on your machine, no API keys are required and prompts do not leave your computer. The server supports server-side conversation replay via sessionId and returns a structured AskResponse containing provider, response, model, sessionId, and usage. It also exposes get-usage-stats and ping. Ollama must be installed and running locally with the target model pulled (for example, ollama pull qwen3.8:27b); if the requested model is not present, the call returns a clear ollama pull error. This is an unofficial third-party tool, not affiliated with, endorsed, or sponsored by Anthropic, Google, OpenAI, or xAI.
Upstash's official server providing up-to-date third-party library docs for AI coding assistants
AI-powered NPM package analysis MCP server
An independent control layer that gives AI coding agents budgets, brakes, and receipts
GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language
Data synced 2026-09-25 Read the FMRS scoring method