Kanwas: Open-source shared context board for teams and AI agents

Kanwas is an open-source shared context board that lets teams and AI agents work on the same canvas simultaneously. Agent tool calls stream into a shared timeline visible to everyone. The project is Apache 2.0 licensed and available on GitHub.

Key features

• Multiplayer workspace: Teams and agents share documents, evidence, and decisions on one canvas.
• Live agent streaming: AI agent tool calls appear in real-time on the same timeline every team member sees.
• Git-backed filesystem: All data is stored as markdown files with full version history.
• Self-hosted: Run with Docker Compose; no vendor lock-in.

Quickstart

Prerequisites: Docker, Docker Compose, and an Anthropic (or OpenAI) API key.

git clone https://github.com/kanwas-ai/kanwas.git
cd kanwas
cp .env.example .env
cp backend/.env.example backend/.env
cp yjs-server/.env.example yjs-server/.env
cp frontend/.env.example frontend/.env
docker-compose --profile app up

Open http://localhost:5173 to start working.

Who it's for

• Founders: Combine fundraising decks, customer interviews, MVP specs, and hiring plans into one canvas with agent assistance.
• Product managers: Drop interview snippets, tickets, and competitor screenshots; get a discovery readout and PRD with traceable claims.
• Developers: Pull PM specs and designs onto a canvas; generate an implementation plan with tasks and acceptance criteria, then push markdown to your repo for Claude Code or Codex.
• Marketers: Plan launches with positioning, messaging, and timelines; draft copy variants side-by-side.
• Sales: Maintain reusable account boards with research, comms history, stakeholder maps, and proposal drafts.

Architecture and stack

Built with TypeScript (98.8%), CSS (0.8%), and Docker. Key dependencies: Yjs (CRDT for real-time collaboration), BlockNote (editor), AdonisJS (backend framework), and E2B (sandboxed code execution). The project uses a monorepo with pnpm workspaces.

Community

Join the Kanwas Kollective on Slack for questions and ideas. Contributors should read docs/SYSTEM_OVERVIEW.md first and open an issue before large changes.