Nanocoder CLI
Nanocoder runs a terminal coding agent on your own model: local via Ollama or any OpenAI-compatible API.
- 2,567 ★ on the repository
- Last commit
Extended access
Why this level
- The agent reads and changes project files, runs commands (without asking in yolo and auto-accept modes) and sends code to the model provider you choose.
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/nanocoderDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The agent reads and changes project files, runs commands (without asking in yolo and auto-accept modes) and sends code to the model provider you choose.
This is third-party code. Review the repository files before installing.
What it does
An open coding agent built by the Nano Collective community. You choose the provider: local models through Ollama or any OpenAI-compatible API such as OpenRouter, Requesty, Anthropic and Google. There are no paid tiers or closed features. Running `nanocoder` opens a fullscreen interface and `nanocoder --provider openrouter --model ... run "task"` runs without a dialog. There are development modes (normal, auto-accept, yolo, plan, architect), branch or PR review with `nanocoder review` and a separate working copy for a PR (`nanocoder worktree 42`). Skills, lifecycle hooks, MCP servers and checkpoints are covered in the docs.
Who it is for. Developers who want a terminal agent not tied to one vendor and able to keep code and the model on their own machine.
Good fit when
- Code must not go to the cloud and the agent needs to run on a local model through Ollama.
- You want to switch providers through OpenRouter without changing the tool.
- A branch or PR needs a check for bugs, vulnerabilities and style with `nanocoder review main`.
- You want to look at someone else's PR in a separate working copy without touching your own: `nanocoder worktree 42`.
Not a fit when
- You need an agent from a large vendor with support guarantees: the project is run by a community.
- The model is weak or local with little memory: quality depends on the chosen model.
- `nanocoder review` must be wired into a pipeline: it works only in an interactive terminal and cannot write output to a file.
Example request
Run nanocoder on a local model through Ollama in plan mode and draw up an audit plan for the auth module without changing files.Limitations
Install through npm, Homebrew or Nix Flakes. Quality and cost depend on the model and provider you choose and requests with your code go to the provider you configured. The `yolo` and `auto-accept` modes run commands without asking, so use them in a safe environment. `nanocoder worktree` needs Git and an authenticated `gh` and leaves the created copy and branch after exit. GitHub shows no license although the project calls itself open, so check the `LICENSE` file.
How to disable. Remove the package with `npm uninstall -g @nanocollective/nanocoder` and the settings folder. Delete PR working copies with `git worktree remove` and `git branch -d`.
FAQ
How do I install it?
`npm install -g @nanocollective/nanocoder`, then `nanocoder`. Homebrew and Nix Flakes are also available.
Which models are supported?
Local ones through Ollama and any OpenAI-compatible API: OpenRouter, Requesty, Anthropic, Google and others.
How do I run it non-interactively?
With `nanocoder --provider openrouter --model google/gemini-3.1-flash run "analyze src/app.ts"`.
What modes exist?
Normal, auto-accept, yolo, plan and architect, selected with `--mode`, for example `nanocoder --mode plan`.
How do I review a PR?
`nanocoder review 42` checks the diff for bugs, security issues and style and `nanocoder worktree 42` opens the PR in a separate working copy.
Discussion
No questions or reviews yet.
Related
CLI toolsA teaching terminal coding agent in Python, built as a "nanoGPT for agents": read it, take it apart and fork it
An open coding agent for the terminal, browser and desktop that works with models from OpenAI and Claude to Ollama
Ollama runs open models on your own machine and serves them to agents through a local API compatible with OpenAI and Anthropic.
A terminal coding agent that reads your project, edits files and runs commands, sending each request to the best free model that is answering today
Agav is a terminal coding agent that reads a repository, edits files, runs tests and switches between seven model providers.
A headless coding agent in one Go binary: reads and edits files, runs commands and never trims context silently
README in short
The README presents Nanocoder as an open community terminal agent with any provider of your choice. It gives a quick start, CLI flags and screen modes.