add-atomic-chat-tool

Yes, you can use local models for private inference within a containerized agent environment. This skill bridges host-side Atomic Chat desktop services with the agent's toolset via the Model Context Protocol, allowing local inference without relying on cloud APIs.

The Model Context Protocol supports local OpenAI-compatible model servers by using standard MCP tool registration within the agent runner. This skill bridges host-side Atomic Chat desktop services to enable local inference for your agent.

To expose local models, ask your agent to list available atomic chat models and then use the atomic_chat_generate tool to process your prompt with a specific local model ID. This tool-based inference is facilitated through standard MCP tool registration.

Yes, Docker is required for local inference tasks with this MCP agent. The skill requires Docker-based network connectivity to bridge host-side Atomic Chat desktop services with the containerized agent environment for private inference.

A limitation of using local LLMs for agent orchestration is the strict requirement for Docker-based network connectivity to bridge the host-side Atomic Chat services. It is designed for offloading private or high-volume tasks while keeping the primary orchestrator remote.