An open-source terminal coding agent, forked from pi-mono (itself derived from Claude Code). It has fewer features than Claude Code by design — the bet is that a small, hackable core you can shape beats a large feature set you can't.
Claude Code is a great product. dreb isn't trying to compete on features — it's trying to compete on flexibility. The core is kept minimal; what you'd find baked into other tools, you build here with skills (markdown workflows), extensions (TypeScript), or install from third-party packages.
Concretely, dreb ships without things Claude Code has — and that's intentional:
- No MCP. Build CLI tools with READMEs (Skills), or build an extension that adds MCP support.
- No permission popups. Run in a container, or build your own confirmation flow with extensions.
- No plan mode. Write plans to files, or build it with extensions, or install a package.
- No background bash in the main agent. The main agent runs commands synchronously. For parallel work, use the subagent tool.
What you get in exchange: a skill system, an extension API, custom agent definitions, custom provider support (route through any proxy, use any API-compatible backend), and a subagent system for parallel work. From those primitives, you build what you need.
git clone https://github.com/aebrer/dreb.git
cd dreb
npm install
npm run build
npm link -w packages/coding-agentThen authenticate and run:
export ANTHROPIC_API_KEY=sk-ant-...
drebOr use a coding subscription (Claude Max, Codex, GitHub Copilot): run dreb then /login.
Or use a custom provider — corporate proxy, Bedrock, local models, anything OpenAI/Anthropic-compatible. See Custom Models.
10 built-in tools: read, write, edit, bash, grep, find, ls, web_search, web_fetch, subagent — plus always-active skill and tasks_update.
mach6 — a built-in development workflow covering the full issue-to-merge lifecycle: assess → plan → implement → push → review → fix → publish. Multi-agent code review with 5 specialized agents running in parallel, independent assessment of findings, iterative review-fix cycles, GitHub as shared memory. Inspired by mach10 by Kevin Ryan.
Skills — markdown workflow definitions the agent loads on-demand. Built-in skills ship with dreb; add your own or install third-party ones.
Extensions — TypeScript modules for custom tools, commands, keyboard shortcuts, event handlers, UI components, custom providers.
Sessions — persistent session tree with branching, compaction, and in-place navigation.
Memory — persistent, file-based memory (global + project-scoped) that survives across sessions.
Custom Providers — route any built-in provider through a proxy, add new providers via JSON config or extensions. 20+ providers supported out of the box.
Modes: Interactive TUI, print/JSON CLI, RPC (for process integration), SDK (for embedding in your own apps).
A hard fork means we control the update cadence. No upstream changes land without us reading the diff first. We cherry-pick what's useful and skip what isn't.
See FORK.md for details.
| Package | Description |
|---|---|
| packages/coding-agent | CLI tool, built-in tools, skills, sessions, extensions — full docs |
| packages/ai | LLM provider abstraction — 20+ adapters, OAuth, model discovery |
| packages/agent | Agent runtime — tool loop, state, streaming, hooks |
| packages/tui | Terminal UI — differential rendering, markdown, syntax highlighting |
MIT