Expose local Ollama models as tools via an MCP server.
npx skills add https://github.com/thmtz/nanoclaw-fleet --skill add-ollama-tool-thmtz
Please help me install this Agent Skill. Skill: add-ollama-tool Source: https://github.com/thmtz/nanoclaw-fleet/tree/main/.claude/skills/add-ollama-tool Command: npx skills add https://github.com/thmtz/nanoclaw-fleet --skill add-ollama-tool-thmtz
This skill enables a stdio-based MCP server inside the agent container to expose local Ollama models as tools, allowing the container-backed agent to call local models and optionally manage the model library.
Enable the Ollama MCP server inside the container and optionally enable management tools, then restart the service.
High-intent search queries and answers about installing and using this skill.
You can expose local Ollama models as tools by running a stdio-based MCP server inside the agent container, enabling the container-backed agent to call local models for low-latency inference.
Yes, you can manage the model library by setting the OLLAMA_ADMIN_TOOLS flag to true, which enables opt-in management tools like ollama_pull_model and ollama_delete_model inside the agent.
Running local inference requires an Ollama runtime on the host, MCP integration in the agent runner, and a container environment to host the stdio-based MCP server.
Using a container agent with an exposed local Ollama model allows you to perform fast, private code tasks locally via the ollama_generate tool without sending data to external services.
The MCP server uses the ollama_list_models tool to list installed Ollama models, returning the model name, size, and family information to the calling agent.
Local Ollama model usage depends on a host Ollama runtime and MCP integration, meaning functionality is limited if the container cannot access the host runtime or if management tools are not explicitly enabled.
VincentChong123Integrate local Ollama models as MCP tools in containerized agents.
OfirGavishExpose local Ollama models as MCP tools for container agents.
clsandovalExpose local Ollama models to the NanoClaw container agent via an MCP server.
NVIDIA-AI-IOTDeploy OpenAI-compatible LLM or VLM serving endpoints on Jetson devices.
