add-ollama-tool

Expose a local Ollama MCP server to container agents for model calls.

Updated Apr 13, 2026

Search Tags:#automation #security #mcp #container #inference #ollama #local-models

Ready to run

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

💬 Prompt for ChatGPT / Claude Chat▼

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

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Expose a local Ollama MCP server to container agents, enabling on-host model calls without leaking host access.

Core Features & Use Cases

Core tools: ollama_list_models and ollama_generate for basic model listing and inference
Optional management tools enabled via OLLAMA_ADMIN_TOOLS for pulling, deleting, showing, and listing running models
Use Case: In a secured container, agents can call local models to generate responses without cloud dependencies

Quick Start

Run the MCP server within the host and verify that ollama_list_models and ollama_generate respond to requests.

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 a local Ollama server to container agents for inference?▼

You can expose local Ollama models to container agents by running a deterministic MCP server that provides tools like ollama_generate, enabling on-host model calls without leaking host access.

Can I manage local models from inside a secured container without cloud dependencies?▼

Yes, by enabling OLLAMA_ADMIN_TOOLS on the MCP server, agents can pull, delete, show, and list running models for offline or secure setups without external cloud dependencies.

What tools are available for Ollama model calls through an MCP server?▼

The MCP server provides core tools ollama_list_models and ollama_generate for basic inference, plus optional management tools for pulling, deleting, showing, and listing running models.

Does this local model MCP setup require giving container agents full host access?▼

No, the setup applies to containerized environments that need in-host model access, allowing agents to call local models through the MCP server without leaking broader host access.

How do I configure optional management tools for an Ollama MCP server?▼

You can configure optional management tools by setting the OLLAMA_ADMIN_TOOLS environment variable, with additional configuration guidance provided through the index.ts and .env files.