add-ollama-tool

Add an Ollama MCP server to NanoClaw for local model calls.

Updated May 7, 2026

Search Tags:#mcp #container #integration-testing #tools #ollama #model-management #host-integration

Ready to run

npx skills add https://github.com/Bartunek/nanoclaw --skill add-ollama-tool-bartunek

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: add-ollama-tool
Source: https://github.com/Bartunek/nanoclaw/tree/main/.claude/skills/add-ollama-tool
Command: npx skills add https://github.com/Bartunek/nanoclaw --skill add-ollama-tool-bartunek

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @modelcontextprotocol/sdk, zod, and includes scripts (resource) components.

What problem does it solve?

This skill exposes Ollama's local model tools via an MCP server to the NanoClaw agent, enabling on-host model invocation and optional model-management.

Core Features & Use Cases

Exposes a local Ollama MCP server inside the containerized NanoClaw agent, bridging host Ollama with the agent's tooling to call local models (list models, generate).
Optional management tools (pull, delete, show, list-running) can be enabled via OLLAMA_ADMIN_TOOLS to manage host models from the agent.
Automatic host-env forwarding wires OLLAMA_HOST and OLLAMA_ADMIN_TOOLS into the container and registers the server for tool discovery.

Quick Start

Frequently Asked Questions about add-ollama-tool

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I expose local Ollama models to a containerized agent as tools?▼

To expose local Ollama models to a containerized agent, you add an Ollama MCP server that bridges host-based models with the agent's tooling, enabling direct model invocation and listing inside the container.

Can I manage Ollama models like pulling and deleting from within my agent?▼

Yes, you can manage Ollama models from within your agent by enabling optional management tools via the OLLAMA_ADMIN_TOOLS environment variable, which exposes pull, delete, show, and list-running functionalities.

How does an MCP server connect a containerized agent to host Ollama instances?▼

An MCP server connects a containerized agent to host Ollama by registering the server inside the container and forwarding host environment variables like OLLAMA_HOST, allowing the agent to discover and call the local models.

Do I need to forward host environment variables to use Ollama with a containerized agent?▼

Yes, you need to forward host environment variables such as OLLAMA_HOST and OLLAMA_ADMIN_TOOLS into the container to properly wire the Ollama MCP server and enable tool discovery and management features.

What dependencies are required to integrate an Ollama MCP server for local model management?▼

Integrating an Ollama MCP server for local model management requires the @modelcontextprotocol/sdk and zod libraries within your scripts to handle server registration and schema validation.