dopejs is a small group of open source projects around one question: what does the software you use every day look like when agents are part of the loop? Agent runtimes, terminal coding agents, CLI environment tooling, Canvas rendering engines — all developed in public.
Five repositories, at honestly different stages of maturity. Status labels below are not marketing — they say what you can actually rely on today. Licensing differs too; each project page states its own.
A personal agent OS — one local daemon, many thin clients.
A Rust control plane runs locally and owns the runtime, LLM providers, channel connectors, storage and events. Clients stay thin: a React web UI, a full-screen Rust TUI, chat-channel connectors, and a TypeScript SDK — all speaking the same JSON Schema contracts.
Lightweight coding agent that runs in your terminal.
A DeepSeek-focused fork of Codex CLI shipping the `dcode` command. Defaults to DeepSeek V4 Flash, supports API-key login, DeepSeek balance and model APIs, and an optional external vision model for image inputs. Signed, checksummed release builds for macOS, Linux and Windows.
curl -fsSL https://github.com/dopejs/dcode/releases/latest/download/install-dcode.sh | sh
Multi-CLI environment switcher with API proxy auto-failover.
Switch environments across Claude Code, Codex and OpenCode from one place, and keep working when an upstream provider degrades — GoZen proxies API traffic and fails over automatically.
curl -fsSL https://raw.githubusercontent.com/dopejs/gozen/main/install.sh | sh
A Web Canvas rendering engine designed from scratch.
A high-performance TSX runtime with native virtual scrolling, a deterministic Rust/WASM core behind a versioned binary ABI, full text rendering and canvas-native editing. Milestones M0–M3 are complete; M4 brings the editing, event, hit-testing and accessibility chain.
An infinite canvas for large collections of AI-generated web artifacts.
Hundreds of live iframes do not scale; flattening every page to an image loses selection and event targeting. dope-canvas keeps artifacts retained — source, durable state, interaction tree, paint cache and an optional live runtime — so Figma-like selection and activation survive. The repository is at its pre-development baseline: architecture, delivery plan and security model, not yet a working canvas.
Everything here is developed in the open, on GitHub. Issues and pull requests are welcome on any repository — and so are honest reports that something does not work yet. Several of these projects are pre-release by design; where that is the case, the project says so.