OSIN.

AN AI CODING ASSISTANT, IN YOUR TERMINAL

Think it through.
Right where you build.

OSIN gives you a focused place to think through coding problems with AI—without leaving your terminal. Bring a compatible API endpoint and model; keep the conversation close to your work.

Get started Explore the source

Terminal-native · Endpoint-configurable · Open source

01 / IN THE TERMINALOSIN, running for real
Screenshot of the real OSIN terminal interface running locally in Windows Terminal, showing its new-chat workspace and the /help command output.
REAL OSIN TERMINAL CAPTURE Captured locally while running /help. The interface and command output are real; no model request was made.
02 / BUILT FOR THE LOOPUseful details, not magic tricks

A calmer place
to work it out.

OSIN is a terminal chat for coding questions. It keeps the interaction simple: choose a model, bring relevant text into the conversation, and stay in your shell.

01⌨

A focused terminal workspace

A mouse-enabled, keyboard-friendly chat interface with multiple conversations, recent chats, and a model settings view.

›Ask a question or describe a coding task.
02⇄

Choose your endpoint

Configure a model name and OpenAI-compatible chat-completions endpoint through environment variables.

OSIN_MODELGROQ_BASE_URL
03▤

Bring text into the chat

Attach UTF-8 text files with /attach, or paste larger text to turn it into an attachment. File and request sizes are bounded.

▤ notes.md READY TO ATTACH
04/

Commands that stay out of the way

Use slash commands for new chats, settings, attachments, undo, copying a response, help, and usage.

/new/attach/usage+ more
05◷

See reported token usage

When the endpoint returns usage data, OSIN totals requests and input/output tokens for the current chat. It is not a cost estimate.

INPUTOUTPUT

OSIN sends chat requests to the configured endpoint. The code does not provide tools that edit project files or run shell commands.

03 / HOW IT WORKSA direct path in

Three steps.
Then get to work.

OSIN is a chat interface, not an autonomous agent. You decide what to ask and what context to share.

  1. 01

    Set your endpoint

    Provide an API key and model. OSIN defaults to Groq’s OpenAI-compatible endpoint, or you can configure another compatible URL.

  2. 02

    Start a conversation

    Launch OSIN in an interactive terminal and describe the coding question you want to work through.

  3. 03

    Add context, as needed

    Attach UTF-8 text files or ask follow-up questions. OSIN sends the chat to your configured endpoint and displays its response.

04 / GETTING STARTEDOne install. Your terminal.
INSTALLATION

From prompt
to first run.

Install OSIN globally, create your private per-user config, add an OpenAI-compatible provider key and endpoint, then launch the workspace. No project files are changed by setup.

  1. 01
    Install globallynpm install -g @kabirprokk/osin-cli
  2. 02
    Initialize your configosin init
  3. 03
    Set key, base URL & modelOSIN_API_KEY · OSIN_BASE_URL · OSIN_MODEL
  4. 04
    List available models, then launch/models · osin
Read the complete installation guide
OSIN / QUICKSTART PUBLIC NPM RELEASE LIVE

# Install globally from npm

npm install -g @kabirprokk/osin-cli

# Create your per-user config

osin init

# Set OSIN_API_KEY, OSIN_BASE_URL and OSIN_MODEL

# In the app: /models lists models exposed by your provider

osin

Requires Node.js 18+. OSIN currently supports OpenAI-compatible Chat Completions APIs. See the README for provider configuration.

Prepared for osin.is-an.org · This page has not been deployed or DNS-verified.

OSIN

Make room for
the next good question.

Explore the source and see if OSIN fits your workflow.

View OSIN on GitHub