openchamber/openchamber

OpenChamber – A unified workspace for AI‑assisted coding

What it is

• An open‑source application that lets you run, supervise, and review code‑writing agents (the OpenCode agents) from any device – desktop app, web/PWA, VS Code extension, or mobile app.
• It stitches together the whole lifecycle of an agent run: launching a task, watching the agent’s output, iterating with the model, reviewing diffs, and finally merging the change back to a repository.

Why it matters

• Modern AI‑coding tools (e.g., GitHub Copilot, OpenCode) are great at generating snippets, but they lack a place to keep long‑running “sessions” organized, to compare multiple model outputs, or to continue work after you close the window. OpenChamber fills that gap with a persistent, multi‑device workspace.

Key capabilities (as described in the README)

Feature	What you can do
Session Goals	Define a finish‑line for a session. The agent keeps working until the goal is met, blocked, or a user‑set limit is reached – even if you close the UI.
Multi‑run & Fusion	Run the same task on up to five models in parallel, each in its own worktree. Pick the best result or automatically merge the strongest parts into a new session.
Changes Walkthrough	Turn a large diff into an AI‑guided, step‑by‑step tour that explains each edit and why it was made.
Preview & Inspect	Open the target app side‑by‑side with the chat, click an element, and send the agent a screenshot, CSS, position, and any browser errors for context‑aware debugging.
GitHub integration	Start a session directly from an issue or PR, feed failed CI checks or review comments to the agent, and merge the PR from within OpenChamber.
Cross‑device continuity	Same projects and sessions are available on Desktop (macOS/Windows/Linux), Web/PWA, VS Code, iOS, Android, and via a CLI/server.
Private Relay	Pair a device with a one‑time QR code; the connection is end‑to‑end encrypted and can be revoked. Supports LAN/VPN, Cloudflare/Ngrok tunnels, and SSH as alternatives.
Dashboard & tracking	See status (running, waiting, finished, failed), approvals, token usage, costs, and organize sessions into folders with notes and reusable actions.
Scheduled work	Run a prompt on a cron‑like schedule (once, daily, weekly, etc.) and optionally attach a Session Goal so the work continues until the goal is satisfied.

Supported surfaces

• Desktop app – native windows, mini‑chat, remote‑machine access, native notifications.
• Web / PWA – runs in any browser, can be installed as a progressive web app.
• VS Code extension – embed sessions next to your code, send selections to the agent, compare runs.
• Mobile (iOS/Android) – review progress, receive alerts, and use a touch‑friendly terminal.
• CLI / Server – headless operation, scheduling, remote‑access management.

Getting started (quick‑start steps from the README)

1. Desktop – download the latest release from the GitHub releases page. On Linux use the provided AppImage (chmod +x … && ./OpenChamber-*.AppImage).
2. VS Code – install the OpenChamber extension from the Marketplace.
3. CLI / Web – install the OpenChamber CLI with the one‑liner script:

   curl -fsSL https://raw.githubusercontent.com/openchamber/openchamber/main/scripts/install.sh | bash
   openchamber --ui-password <your‑password>
   

   Then use commands like openchamber status, openchamber connect-url --qr, openchamber tunnel start …, etc.
4. Follow the linked guides (quick‑start, installation, multi‑run, session goals, etc.) for deeper workflows.

Typical use cases

• Automated feature implementation – define a goal like “add pagination to the user list”, let the agent iterate until the UI works and tests pass, then merge the PR.
• Bug‑fix triage – start a session from a failing CI check, feed the error logs to the agent, and have it propose a fix that you can review via the Changes Walkthrough.
• Exploratory prototyping – spin up several models with the same prompt, compare their code, and fuse the best parts into a new prototype.
• Remote pair‑programming – a developer on a laptop can hand off a running session to a teammate on a phone, preserving all context and progress.

Technology stack (as implied by the README)

• OpenCode – the underlying LLM‑powered coding agent platform (https://opencode.ai).
• Electron (for the Desktop app) and Node.js 22+ for the CLI/web server.
• VS Code extension API, PWA technologies, and native mobile wrappers for iOS/Android.
• Optional tunneling providers (Cloudflare, Ngrok) and standard SSH/LAN for remote access.

License

• MIT – free for personal, academic, or commercial use.

---

Bottom line: OpenChamber is a full‑featured, cross‑platform environment that lets developers keep AI‑driven coding work organized, observable, and under control from any device. It is especially useful for teams that want to harness LLM coding agents while retaining a reviewable, auditable workflow.