skills.rest
RADHA0-max
RADHA0-maxCOMMUNITY
Explore all skills by this author →
GitHub

add-ollama-tool

Registers an Ollama MCP server exposing local model inference and library management tools to container agents.

2|Updated Jan 14, 2026
Search Tags:#typescript#mcp server#model management#ollama#local models#llm inference#container agent
One-click install
npx skills add https://github.com/RADHA0-max/selfhealbackend --skill add-ollama-tool-radha0-max
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: add-ollama-tool
Source: https://github.com/RADHA0-max/selfhealbackend/tree/main/nanoclaw/.claude/skills/add-ollama-tool
Command: npx skills add https://github.com/RADHA0-max/selfhealbackend --skill add-ollama-tool-radha0-max

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @modelcontextprotocol/sdk, zod, typescript, vitest.

What problem does it solve? Containerized agents cannot reach local Ollama models by default, forcing all inference through the orchestrator model. This Skill wires a stdio-based MCP server into the NanoClaw agent-runner so the agent can offload work to local models and optionally manage the Ollama model library. ## Core Features & Use Cases - Local model inference tools: Exposes ollama_list_models and ollama_generate so the agent can list installed models and send prompts to them via the Ollama REST API. - Opt-in library management: Setting OLLAMA_ADMIN_TOOLS=true adds tools to pull, delete, inspect, and list running models. - Guided installation with wiring tests: Copies the MCP server source into the container tree, registers it in index.ts, forwards OLLAMA_HOST env vars, and validates both integration points with structural tests. - Use Case: You want your container agent to summarize long documents cheaply. After applying this Skill, the agent calls ollama_generate with a local model like gemma3:1b instead of consuming orchestrator tokens. ## Quick Start Ask the agent to add the Ollama tool so it can call local models, then confirm whether you want model-management tools enabled.

Frequently Asked Questions about add-ollama-tool

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

FAQPage Schema
How do I let a container agent use local Ollama models?▼

Register a stdio MCP server that wraps the Ollama REST API and expose its tools to the agent. This Skill copies the server into the agent-runner tree, registers it in index.ts, and the agent gains ollama_list_models and ollama_generate tools.

How do I enable Ollama model management tools for an agent?▼

Set OLLAMA_ADMIN_TOOLS=true in your .env file and restart the service. This opt-in flag registers ollama_pull_model, ollama_delete_model, ollama_show_model, and ollama_list_running alongside the default list and generate tools.

Does the Ollama MCP server need API keys or credentials?▼

No, Ollama runs locally and is keyless. The only configuration is OLLAMA_HOST, the daemon's base URL, which defaults to http://host.docker.internal:11434 with a localhost fallback.

Why does the agent say Ollama is not installed or try to run a CLI?▼

The agent is looking for an ollama CLI inside the container instead of using MCP tools. Verify the server file exists in container/agent-runner/src, the ollama entry is registered in the mcpServers object in index.ts, and the container image was rebuilt.

Why does ollama_generate return model not found errors?▼

The model name must exactly match a name returned by ollama_list_models, including any tag suffix like gemma3:1b. Have the agent list models first and pick one from that list before generating.

What are the limitations of ollama_pull_model for large models?▼

The pull tool uses stream:false and blocks until the download completes, so large models over 7B can take several minutes. For very large pulls, run ollama pull directly on the host CLI instead.

Related Skills

Explore More Software Engineering Skills→
qwibitaiqwibitai
10

add-ollama-tool

Expose local Ollama models as MCP server tools for summarization and queries.

Official
Intermediate
clsandovalclsandoval

add-ollama-tool

Expose local Ollama models to the NanoClaw container agent via an MCP server.

Community
Intermediate
jboothomasjboothomas
1

add-ollama-tool

Integrate local Ollama models into container agents via MCP tools.

Community
Advanced
understudylabsunderstudylabs
10

run-local-model-lab

Compare local MLX models against frontier gateways on latency, cost, and quality.

Official
Advanced