ChatOllama

ChatOllama is an AI agent app whose main interface is a command-line client running on your machine. It's for people who want an agent to work with local files while choosing between models served by Ollama and hosted services. The repository license modifies Apache 2.0: commercial hosting and multi-tenant services require authorization, and frontend branding must remain intact.

The client connects to Ollama through an OpenAI-compatible API, or to OpenAI, Anthropic, Google Gemini, DeepSeek and OpenRouter. It discovers installed Ollama models and models available through configured remote providers, then lets you switch models within the client. Hosted providers require API credentials and send model requests to their services; with Ollama, requests go to the Ollama endpoint you choose.

Agents can inspect directories, read files and search workspace contents. They can also write files and make exact text edits. File access stays within the workspace directory, including checks that reject paths or symlinks leading outside it. Edits can detect stale file contents before overwriting them, and the workspace tools don't expose shell commands.

The terminal shows tool inputs, execution status and bounded results, so you can see what the agent is doing with your files. Follow-up prompts share the same in-memory conversation, and you can cancel a response while keeping the client open. ChatOllama remembers your selected model for the next session, but doesn't save conversations or API keys. The client requires Node.js 24 or newer.