This is an early release preview. You may encounter bugs.
Nanocoder logo
Unclaimed

Tool coding

Nanocoder

An open coding agent for your terminal, built by a community collective rather than a company. Bring your own model, keep your code on your

A 89/100 GitHub score ? This grade is derived from GitHub signals, not user votes. Open for the full breakdown.
No votes yet

01 / About

What Nanocoder is.

Nanocoder is a terminal coding agent that you point at a model of your choosing: local models through Ollama, or any OpenAI-compatible API, including OpenRouter, Anthropic, and Google endpoints. Code stays on your machine, and the provider decision, and therefore where your data goes, is yours. It installs from npm, Homebrew, or a Nix flake.

The command line accepts provider, model, and starting mode directly, so a single invocation can run non-interactively against a named provider and model, or open an interactive session in a chosen development mode: normal, auto-accept, yolo, or plan.

Two rendering modes are available. The default inline mode prints finished messages into the terminal's native scrollback, so the scrollbar, mouse wheel, and search keep working and the transcript survives after you exit. Fullscreen mode, enabled with --alt-screen or a preference, uses a fixed-height layout on the alternate screen buffer with in-app scrolling, a scroll indicator, and snap-back to the bottom; because mouse reporting takes click-drag selection from the terminal, Ctrl+P toggles selection mode back. In both modes /clear resets to a fresh welcome banner and exiting removes the input UI cleanly.

Documented capabilities beyond the chat loop include skills covering commands, subagents, tools, and event triggers, a per-project daemon, checkpointing, task management, and MCP server configuration, along with logging and timeout settings. The project is built by the Nano Collective as a community effort, with no paid tiers, no telemetry of prompts, and multi-provider support as a stated design constraint.

Features

  • Bring your own model: local Ollama models or any OpenAI-compatible API, selected per run
  • CLI flags: provider, model, and starting mode set directly, before or after the run command
  • Development modes: normal, auto-accept, yolo, and plan, selectable at launch
  • Inline rendering: output lands in native terminal scrollback and remains after exit
  • Fullscreen mode: alternate-screen layout with in-app scrolling and a selection-mode toggle
  • Skills: commands, subagents, tools, and event triggers as reusable units
  • Per-project daemon: background process scoped to the project you are working in
  • Checkpointing: saved states for longer sessions
  • MCP servers: configurable external tool servers
  • Multiple install paths: npm, Homebrew, and Nix flakes

02 / Discussion CREDIBILITY-GATED

Discussion

Reading is open to everyone. Posting and voting need a verified identity or a GitHub grade of B or higher.

  • No discussions yet.

04 / Build

Build with Nanocoder.

Browse the catalogue for frameworks, tools, and harnesses, each scored on real GitHub credibility.

Get Nanocoder →

Browse the catalogue