Claude Sidekick
Provides tools to generate text, chat, embed text, generate code, summarise text, list models, and pull models using locally running Ollama models.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Claude Sidekick generate a simple email template for customer onboarding using Ollama"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Claude Sidekick
A Model Context Protocol (MCP) server that connects Claude to your local Ollama models, allowing you to offload simpler tasks and save your Claude tokens for complex work.
Quick Start
1. Install Ollama
First, install Ollama on your system:
# macOS
brew install ollama
# Linux
curl -fsSL https://ollama.ai/install.sh | sh
# Windows
# Download from https://ollama.ai/downloadStart Ollama:
ollama serve2. Pull Some Models
Download useful models for different tasks:
# General purpose models
ollama pull gpt-oss # OpenAI's open-weight model
ollama pull llama3.2 # Fast, capable model
ollama pull qwen2.5 # High-quality text generation
# Specialised models
ollama pull deepseek-coder # Code generation
ollama pull nomic-embed-text # Text embeddings
ollama pull llama3.2:1b # Lightweight for simple tasks3. Set Up the MCP Server
Create a new directory and install dependencies:
mkdir mcp-ollama-server
cd mcp-ollama-server
# Copy the files (index.ts, package.json, tsconfig.json)
# Then install dependencies:
npm install
# Create src directory and move index.ts there
mkdir src
mv index.ts src/
# Build the project
npm run build4. Configure Claude Desktop
Add the server to your Claude Desktop configuration:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"ollama": {
"command": "node",
"args": ["/absolute/path/to/your/mcp-ollama-server/dist/index.js"],
"env": {}
}
}
}5. Restart Claude Desktop
Restart Claude Desktop to load the new MCP server. You should see the Ollama tools available in Claude.
Related MCP server: LMStudio-MCP
Available Tools
ollama_generate_text
Generate text for simple writing tasks, basic summaries, or straightforward content creation.
Best for: Simple writing, basic explanations, content generation
ollama_chat
Have conversations with local models for Q&A, explanations, or dialogue-based tasks.
Best for: Q&A sessions, explanations, interactive tasks
ollama_embed_text
Generate text embeddings for semantic similarity, clustering, or search.
Best for: Document similarity, semantic search, clustering
ollama_code_generation
Generate code using specialised coding models.
Best for: Simple scripts, boilerplate code, basic programming tasks
ollama_summarise
Summarise text content with different length options.
Best for: Document summaries, article condensation
ollama_list_models
List all available models on your Ollama installation.
ollama_pull_model
Download new models to Ollama.
Usage Examples
Once configured, Claude can use these tools like this:
Text Generation:
"Use Ollama to generate a simple email template for customer onboarding"
Code Generation:
"Have DeepSeek Coder create a Python script to parse CSV files"
Embeddings:
"Generate embeddings for these document titles using Nomic"
Summarisation:
"Use Llama to create a brief summary of this article"
Model Recommendations
For Text Generation
gpt-oss - Excellent balance of speed and quality
llama3.2 - Good balance of speed and quality
qwen2.5 - Higher quality but slower
llama3.2:1b - Very fast for simple tasks
For Coding
gpt-oss - Excellent code generation
deepseek-coder - Great code generation
qwen2.5-coder - Alternative coding model
For Embeddings
nomic-embed-text - Fast, high-quality embeddings
mxbai-embed-large - Larger embedding model
Configuration
Customising the Server
Edit src/index.ts to modify:
Base URL: Change Ollama endpoint (default:
http://localhost:11434)Timeout: Adjust request timeout (default: 5 minutes)
Default Models: Modify which models are used by default
Temperature Settings: Adjust creativity/randomness
Environment Variables
You can override settings with environment variables:
export OLLAMA_BASE_URL=http://localhost:11434
export OLLAMA_TIMEOUT=300000Troubleshooting
"Failed to connect to Ollama"
Ensure Ollama is running:
ollama serveCheck if it's accessible:
curl http://localhost:11434/api/tagsVerify no firewall blocking port 11434
"Model not found"
List available models:
ollama listPull the missing model:
ollama pull model-name
"MCP server not appearing in Claude"
Check the config file path is correct
Verify the absolute path to the built JavaScript file
Restart Claude Desktop completely
Check Claude Desktop logs for errors
Performance Tips
Use smaller models for simple tasks:
llama3.2:1bfor basic text generationqwen2.5:0.5bfor very simple tasks
Keep frequently used models warm:
# Pre-load models to keep them in memory ollama run llama3.2 "hello" ollama run deepseek-coder "print hello"Adjust temperature based on task:
Low (0.1-0.3) for factual/coding tasks
Medium (0.5-0.8) for creative writing
High (0.9-1.2) for very creative tasks
Development
Running in Development
npm run dev # Run with hot reload
npm run watch # Watch mode
npm run type-check # Check TypeScript types
npm run lint # Lint codeAdding New Tools
To add new capabilities:
Define the tool in
setupHandlers()Add the handler method
Update the Ollama client if needed
Rebuild and restart
Extending Model Support
To support new model types:
Add model configuration in the constructor
Create specific handler methods
Add appropriate prompting strategies
Example Workflow
Here's how you might use this in practice:
Initial analysis with Claude: "I need to analyse this dataset and create a comprehensive report"
Delegate simple tasks: "Use Ollama to generate basic descriptions for each data column"
Complex analysis with Claude: Claude does the sophisticated statistical analysis and insights
Offload summarisation: "Use Llama to summarise each section of findings"
Final review with Claude: Claude assembles everything into a polished report
This approach maximises your Claude token efficiency while still getting comprehensive results.
Security Notes
This server runs locally and doesn't send data externally
All model inference happens on your machine
No API keys or external services required
Your data stays completely private
Contributing
Feel free to extend this server with additional capabilities:
Image generation support
Model fine-tuning integration
Performance monitoring
Model switching strategies
Custom prompt templates
Available Tools
7 toolsollama_chatA
Have a conversation with local Ollama for SIMPLE Q&A, factual questions, or basic explanations that don't require deep reasoning. Prefer for routine queries to save Claude tokens. AVOID for complex analysis, nuanced discussions, or tasks requiring sophisticated reasoning.
Inputs: messages (required), model, temperature.
ollama_code_generationA
Generate SIMPLE code like getters/setters, basic CRUD operations, validation rules, boilerplate code, or routine functions. Use for mechanical coding tasks that follow established patterns. AVOID for architectural decisions, complex business logic, or code requiring sophisticated design patterns.
Inputs: task (required), model, language, temperature.
ollama_embed_textA
Generate text embeddings using local embedding models like nomic-embed-text. Ideal for batch embedding tasks, semantic search, similarity comparisons, and clustering. Use this for routine embedding generation to save Claude tokens.
Inputs: text (required), model.
ollama_generate_textA
Generate text using local Ollama for SIMPLE, token-efficient tasks like basic content, error messages, placeholder text, boilerplate code, or routine documentation. Use instead of Claude for non-analytical text generation. AVOID for complex reasoning, analysis, or creative writing that requires nuanced understanding.
Inputs: prompt (required), model, max_tokens, temperature.
ollama_list_modelsA
List all available Ollama models on the local system
Inputs: no named parameters.
ollama_pull_modelA
Download and install a new model to Ollama
Inputs: model (required).
ollama_summariseA
Create BRIEF summaries for logs, documentation, or simple content. Use for factual condensation and routine document processing that doesn't require deep analysis or insight. Ideal for batch summarization tasks to save Claude tokens. AVOID for content requiring interpretation or analytical summarization.
Inputs: text (required), model, length.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 7 tools
The four text-generation tools (generate_text, chat, code_generation, summarise) have overlapping purposes and require careful reading of descriptions to distinguish. Embeddings and model management are clearly distinct.
Naming mixes verb_noun (generate_text, embed_text, list_models, pull_model), bare verbs (chat, summarise), and noun_noun (code_generation). The British spelling of 'summarise' further breaks consistency.
Seven tools is a well-scoped count for an Ollama integration, covering generation, chat, embeddings, and model management without unnecessary bloat.
Core operations are covered (generate, chat, embed, list, pull), but missing model deletion and model info endpoints are minor gaps that agents could work around.
Related MCP Connectors
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
Talk to your public-facing AI from any MCP client — Claude, ChatGPT, Cursor, Cline, Windsurf.
Shared memory MCP server for Claude Code, Codex, Gemini CLI and Cursor. Cited, editable.
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
Related MCP Servers
- AlicenseBqualityDmaintenanceA Model Context Protocol (MCP) server that lets you seamlessly use OpenAI's models right from Claude.1319 npm77MIT
- AlicenseNot gradedqualityDmaintenanceA Model Control Protocol (MCP) server that allows Claude to communicate with locally running LLM models via LM Studio.MIT
- FlicenseNot gradedqualityDmaintenanceMCP server that lets Claude Code offload simple tasks like code explanation, writing tests, and adding comments to a local Ollama model, saving Claude API tokens.-
- AlicenseNot gradedqualityDmaintenanceAn MCP server that allows Claude Code to offload mechanical tasks such as summarization, classification, and drafting to a local LLM, reducing API costs while keeping Claude in control of complex reasoning and quality review.MIT