add-ollama-tool

Name: add-ollama-tool
Availability: InStock
Author: jboothomas

Integrate local Ollama models into container agents via MCP tools.

1|2|Updated Apr 8, 2026

Search Tags:#tooling #mcp #ollama #model-management #local-models #container-agent #auggie

Ready to run

npx skills add https://github.com/jboothomas/nauggieclaw --skill add-ollama-tool-jboothomas

💬 Prompt for ChatGPT / Claude Chat▼

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

NauggieClaw container agents lack direct access to local Ollama models and manual setup is fragile, so this skill wires in an MCP server that lets the agent call and optionally manage those local tools securely.

Core Features & Use Cases

Local tool bridge: Adds ollama_list_models and ollama_generate so the agent can inspect installed Ollama models and run inference without exposing the CLI inside the container.
Optional management: Activating OLLAMA_ADMIN_TOOLS=true lets the agent pull new models, delete old ones, show detailed metadata, and list running instances, keeping the model library synchronized via prompts.
Deployment workflow: Walks through verifying prerequisites, merging the upstream skill branch, copying updated files to cached runners, rebuilding the container, configuring .env, restarting the service, and verifying by sending inference and management requests.

Quick Start

Tell Auggie to use Ollama to tell me the capital of France.

Frequently Asked Questions about add-ollama-tool

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

FAQPage Schema

How do I connect container agents to local Ollama models for inference?▼

Connecting container agents to local Ollama models requires an MCP server bridge that exposes tools like ollama_generate, letting agents run inference on host Ollama servers without CLI access inside the container.

Can my agent pull and delete Ollama models automatically to keep the library synchronized?▼

Yes, agents can pull, delete, and inspect Ollama models automatically by enabling the OLLAMA_ADMIN_TOOLS configuration toggle, which keeps the local model library synchronized through prompt-driven management.

What is the best way to list running Ollama instances from a containerized agent?▼

Listing running Ollama instances from a containerized agent is best handled by an MCP integration providing ollama_list_models, securely querying the OLLAMA_HOST to return installed and active model statuses.

Do I need to expose the Ollama CLI inside my container to run local models?▼

No, you do not need to expose the Ollama CLI inside the container, because an MCP server integration allows the agent to call ollama_generate and manage models securely over the host connection.

How to verify Ollama MCP tools are working after rebuilding a container agent?▼

Verify Ollama MCP tools after rebuilding by configuring the .env file with OLLAMA_HOST, restarting the service, and sending test inference or management requests to confirm the container agent receives valid responses.

Why does my container agent fail to call local Ollama models?▼

Container agents fail to call local Ollama models when the MCP server bridge is missing or the OLLAMA_HOST configuration is incorrect, preventing the agent from reaching the host server for inference tasks.