Skip to content
Context Code Context Code
Context Code

Context Code

Create without limits. The local-first coding assistant that chats, edits, orchestrates agents and runs models on your own machine.
Install Context Code Your first chat
Context Code in ninety seconds, narrated — everything you see is a real screenshot of the running application.

What is Context Code?

Section titled “What is Context Code?”

Context Code is a desktop app (Windows, macOS and Linux) where you work with AI agents on your own projects and on your own machine: chats with real access to your files, a code editor with LSP, a terminal, a Git panel, an SQL console, orchestration of agent teams and an embedded local model engine (Context Forge). All your data — chats, history, settings — is stored encrypted on your disk, never on a Context Code server.

The main window: chat with the agent on the right, file explorer and editor in the center.

Your provider, your choice

Anthropic (Claude), OpenAI, Google Gemini, GitHub Copilot, xKiro, OpenRouter, DeepSeek and more — or 100% local models with Ollama, LM Studio and Context Forge. Switch models on the fly, inside the same chat.

Talk to it

Real-time voice conversation: it listens while you speak, you can interrupt it and it answers with the voice you choose. The Context Voice widget floats always on top and understands commands for the app itself.

Agents that do, not just talk

The agent reads and edits files, runs commands and navigates your code through a knowledge graph. You decide how much power to give it with the approval modes.

From wherever you are

A mobile app paired by QR, web access from your phone’s browser, and WhatsApp, Telegram and Slack channels to talk to your agent without sitting at the PC.

Private by design

Database encrypted with SQLCipher, keys in the system keychain, no telemetry. Remote channels ship turned off until you turn them on.

A 30-second tour

Section titled “A 30-second tour”
  1. Install the app and open it: Installation.
  2. Connect an AI provider (or download a local model): Providers.
  3. Open your project and type your first request in the chat: Your first chat.
  4. Decide how much you trust the agent with the approval modes: Approval modes.
  5. When you want more: orchestrate a team of agents, schedule automations, talk to it in real time or pair your phone.

Coming from an earlier version? See what’s new in 3.6.