Code you can read, run, and fork — UniCore, LiteDuck, and the BWOC framework with its plugins, MCP server, gateway, and apps.
Lightweight native code editor — no AI, no clutter
A fast, lightweight native code editor — file browser, editor, integrated terminal, and Git in one focused desktop app. No AI, no clutter.
A CLI and web console that builds AGI agents you own
A CLI and web console that builds AGI agents you own — a soul of its own, memory recalled from your notes and documents, autonomy with a kill switch, A2A and Telegram, one git repo per agent.
The code is public and free to use under the Business Source License, which limits some production use — it is not an OSI open-source license. Check the license before you deploy.
Open framework for managing AI agent fleets
Declare a whole agent fleet, drive it from your editor, and keep every agent backend-neutral. Declarative, shell-out, no daemon.
Drive the BWOC fleet from Claude Code
BWOC → Claude Code plugin adapter — coordination, agents, skills, and deep-memory wired into Claude Code via the bwoc CLI.
Drive the BWOC fleet from OpenAI Codex
BWOC → OpenAI Codex plugin adapter — coordination, skills, and deep-memory via the bwoc CLI.
Native Cursor plugin for the BWOC fleet
Native Cursor editor plugin — /bwoc-* commands, a fleet rule and skill, and sub-agents, all wrapping the bwoc CLI.
Drive the BWOC fleet from Antigravity
BWOC → Antigravity plugin adapter — coordination, skills, rules, and deep-memory via the bwoc CLI.
Drive the BWOC fleet from Hermes
BWOC → Hermes plugin adapter — tools, skills, and a memory provider via the bwoc CLI.
Drive the BWOC fleet from OpenClaw
BWOC → OpenClaw plugin adapter — coordination tools, agents, skills, and a memory slot via the bwoc CLI.
Universal BWOC fleet plugin for AI coding agents
A universal BWOC fleet plugin for AI coding agents — Claude Code, Codex, Cursor, Copilot, and Grok — installable with one command.
Expose a BWOC workspace to any MCP client
A public MCP server for BWOC — expose agents, teams, tasks, memory, and run/send/fleet to any MCP client over stdio or Streamable HTTP.
Relay server for multi-place agent communication
A rendezvous and relay server for multi-place agent communication — a signed-envelope WebSocket relay.
Browser control surface for the agent fleet
An MV3 Chrome extension that drives the BWOC daemon (bwocd) over ed25519-signed requests — a browser control surface for the fleet.
macOS menu-bar control for the fleet
A SwiftUI macOS menu-bar control center for the BWOC agent fleet — glance and act without leaving the menu bar.
Native desktop chat for BWOC agents
Native desktop chat for BWOC agents — talk to one agent, or a whole team, in a single window. Built with egui.
Guided setup wizard for the BWOC framework
A non-coder-friendly installer with a guided TUI setup wizard (bwoc-setup) for the BWOC framework.
Bilingual guide to the BWOC framework
A bilingual (EN/TH), role-indexed guide to the BWOC framework — for end-users, developers, agent authors, AI search, and crawlers.
Voice in and out for BWOC agents
Voice (TTS) and voice-input (STT) for BWOC agents — on-device F5-TTS-MLX or remote F5-TTS-THAI, per-agent voices, and push-to-talk Whisper.
Dashboard over the bwoc CLI
A dashboard over the bwoc CLI for BWOC agent fleets.
Operate your fleet from VS Code
Operate your BWOC agent fleet from VS Code — a native control surface over the bwoc CLI (read-only fleet view in the first release).
BWOC treats a fleet of AI agents like a codebase — declarative, versioned, and verifiable — instead of a pile of prompts.
Describe every agent — its role, skills, and routes — in plain config. Incarnate a new one with a single command and the framework scaffolds the rest.
Drive the whole fleet from your terminal or editor with the bwoc CLI: list, status, check, and task — no dashboard to babysit.
One AGENTS.md is the single source of truth; per-backend files are symlinks. A neutrality check keeps agents portable across Claude, Codex, Kimi, and more.
No long-running service, no hidden state. BWOC shells out to the tools you already run, so it stays transparent, scriptable, and easy to audit.
Coordinate work across agents and teams with a file-backed task queue and explicit interconnect routes — peer agents stay loosely coupled.
Multi-agent work runs in isolated git worktrees, so parallel agents never clobber each other and every change stays reviewable.
From a single agent to a coordinated team in three moves.
Run bwoc new <name> to scaffold an agent, then fill in its role and skills. The framework wires up its instruction files and registration.
Group agents into teams, declare interconnect routes, and let them hand tasks to one another through the shared queue.
Run the fleet from your editor. A neutrality check and verification gates run on every change, so agents stay correct and portable.
BWOC is the framework we run our own agent fleet on. It's open — read the source, run it yourself, or have us stand it up for your team.