kanbu
Self-hosted project management for OpenClaw users. Dispatch tasks to AI agents, track their work, and stay in control.
Install / Use
claude mcp add hydro13 -- npx -y github:hydro13/kanbuIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
OperationsSupported Platforms
Our assessment of kanbu
kanbu scores 77/100 on our quality scale, 688th of 758 Operations skills we index.
Its MCP Server is 15 KB long, well organised into 37 sections with 6 code examples: a thorough specification that gives an agent plenty to work with.
It has 10 GitHub stars, so there is little community track record yet; judge it on its content.
Maintenance, license and trust
- The repository was last updated about 8 months ago. That is recent enough to be usable, but agent tooling moves fast, so check the instructions against your agent's current version.
- Our last check on 2026-09-17 found the source still online.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 91/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
kanbu compared with similar skills
All 4 of these similar skills score higher than kanbu; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| kanbu (this skill)by hydro13 | 77 | 10 | 8mo ago | MCP Server |
| taste-skillby Leonxlnx | 100 | 108.6k | 2d ago | CLAUDE.md |
| Agent-Reachby Panniantong | 100 | 108.0k | 1d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 78.7k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 47.8k | today | CLAUDE.md |
Frequently asked questions
- How do I install kanbu?
- Run
claude mcp add hydro13 -- npx -y github:hydro13/kanbu. The install tabs above show the steps for each supported agent. - Which AI agents does kanbu work with?
- It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
- Is kanbu safe to use?
- It is MIT-licensed and scores 91/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is kanbu still maintained?
- The repository was last updated about 8 months ago. That is recent enough to be usable, but agent tooling moves fast, so check the instructions against your agent's current version.
Skill content
View source on GitHubKanbu
Self-hosted project management for people who work with AI agents.
<img width="2428" height="1846" alt="Kanbu Board" src="https://github.com/user-attachments/assets/6cd89fcd-2afa-44c1-a200-8cef163e8c31" />Built for people who work with AI agents
I built Kanbu because I use OpenClaw every day. My projects regularly cross 300,000 lines of code — and at that scale, you need somewhere to track what your agents are doing, what's been decided, and what comes next. No tool did that. So I built one.
Kanbu connects directly to your local OpenClaw gateway. Dispatch any task to an agent from the task view, read its response, steer it with follow-up instructions, and let it update the board as it works. Your task list becomes the shared memory between you and your agents.
Works with any MCP-compatible agent — OpenClaw, Claude Code, Cursor, or any tool that speaks MCP.
It's MIT. It's self-hosted. It's here because I needed it, and maybe you do too.
Why Kanbu?
Most project management tools are either too simple (Trello) or too complex (Jira). And none of them understand AI-native workflows.
Kanbu is different:
| Problem | Kanbu Solution | | -------------------------------------------------------------- | -------------------------------------------------------------------- | | "I need fine-grained permissions but Trello doesn't have them" | Scoped ACL with inheritance, deny rules, and security groups | | "I want AI to help but it can't access my tasks" | 154 MCP tools - Claude Code works directly in your board | | "I want to dispatch tasks to AI agents and see what they do" | OpenClaw agent dispatch — one click, agent runs, response logged | | "GitHub issues and my PM tool are never in sync" | Bi-directional GitHub sync with webhook integration | | "I want to self-host but lose features" | No seat fees, no feature gates — everything included, always | | "My backups are not secure or automated" | Automated backups with AES-256-GCM encryption & scheduling | | "My team speaks different languages" | Each user talks to their own AI assistant in their own language |
Key Features
🎯 Project Management
- Kanban boards with drag-and-drop, swimlanes, and WIP limits
- Multiple views: Board, List, Calendar, Timeline
- Sprints & Milestones with burndown charts
- Time tracking with estimates vs. actuals
- Real-time collaboration: Live cursors, typing indicators, and instant sync via Socket.io
🔐 Permissions & Access Control
Kanbu implements a permission system inspired by filesystem ACLs (think Windows NTFS / Active Directory, but for a web app).
- Granular permissions: Read, Write, Execute, Delete, Permissions — each independently grantable or deniable.
- Inheritance: Workspace → Project → Task permissions flow down automatically.
- Deny-first logic: Explicit deny overrides any grant (e.g., ban a specific user from a project even if they are an admin).
- Audit logs: Complete security trail with export to CSV/JSON.
🧠 Built-in Knowledge Graph Engine
Kanbu includes a wiki intelligence system built in TypeScript (33,000+ lines).
The base wiki runs without any extra services and includes:
- BM25 full-text search: Fast keyword search across all wiki pages
- Contradiction detection: Automatically flags conflicting information across your documentation
- Community clustering: Groups related knowledge using label propagation (own algorithm)
- Version history: Up to 20 versions per page, with restore
The full graph engine (optional, requires Python/FalkorDB via docker-compose.openclaw.yml) adds:
- Semantic search: Find knowledge using vector embeddings, not just keywords
- Temporal queries: Ask "What did we know about this feature last month?"
- Duplicate deduplication: Detects and merges overlapping knowledge nodes across pages
- RAG: Retrieval-Augmented Generation — AI answers grounded in your actual wiki content
- D3.js graph visualization: Interactive visual map of your knowledge graph, live in the browser
Inspired by Graphiti — thanks for the foundation.
🤖 AI Agent Dispatch (OpenClaw)
If you run OpenClaw, Kanbu connects directly to your local agent gateway. Assign any task to an AI agent with one click — no copy-pasting, no context switching.
- One-click dispatch: Open any task → Agent tab → select agent → Dispatch
- Task context included automatically: Title, description, project and workspace name are sent as a structured prompt
- Custom instructions per run: Add extra steering without editing the task
- Run history with response: Every run is logged with status, duration, and the agent's full response
- Iterative control loop: Dispatch → read response → dispatch again with new instructions
- Fire-and-forget: Returns immediately, status updates async as the agent works
Setup (2 env vars):
OPENCLAW_GATEWAY_URL=http://127.0.0.1:18789
OPENCLAW_GATEWAY_TOKEN=your-token-here
When not configured, the Agent tab shows a friendly "not configured" state instead of an error.
<img width="2852" height="1722" alt="Kanbu Agent Dispatch" src="https://github.com/user-attachments/assets/ea98dd79-2d19-46b8-b686-1040a9cdc5b4" />🐾 OpenClaw Skill
Kanbu ships with a ready-made OpenClaw skill. Install it once, and your OpenClaw agent knows how to manage projects, tasks, wikis, and teams — no manual configuration needed.
# Copy the skill into your OpenClaw workspace
cp -r packages/openclaw-skill ~/.openclaw/workspace/skills/kanbu
# Restart your gateway
openclaw gateway restart
That's it. Your agent can now create tasks, move cards, add comments, search across projects, run analytics, and manage your entire board — all through natural language.
🔌 AI Agent Integration (MCP)
Connect any MCP-compatible AI agent to manage your Kanbu projects with 154 available tools.
Tool Categories:
| Phase | Category | Tools | Examples |
| --------- | ------------------- | ----- | ------------------------------------------------------------------- |
| 1 | Pairing | 3 | kanbu_connect, kanbu_whoami, kanbu_disconnect |
| 2 | Core | 11 | kanbu_list_workspaces, kanbu_create_task |
| 3 | Subtasks & Comments | 9 | kanbu_create_subtask, kanbu_add_comment |
| 4 | Search & Activity | 5 | kanbu_search_tasks, kanbu_recent_activity |
| 5 | Analytics | 4 | kanbu_project_stats, kanbu_velocity |
| 6-12 | Admin & Settings | 90 | User management, ACL, Audit logs, Backups, Profile |
| 9+ | GitHub Connector | 10 | kanbu_list_github_prs, kanbu_create_github_branch |
| 17 | Wiki Management | 18 | kanbu_list_project_wiki_pages, kanbu_create_workspace_wiki_page |
| 13-16 | Hardening & Audit | 4 | Rate limiting, Security hardening |
🐙 GitHub Integration
- Bi-directional sync: Issues, PRs, commits, milestones
- Auto-link commits: Reference tasks in commit messages
- Sync logs: Complete visibility into what synced and when
Requires a GitHub App installed on your account or organization. Setup guide →
<img width="2300" height="1407" alt="GitHub Integration" src="https://github.com/user-attachments/assets/154b6f4f-d23f-4dcf-88db-6b374b168b4c" />💾 Backup System
Complete backup solution with strong encryption:
- AES-256-GCM encryption at rest (optional, via
BACKUP_ENCRYPTION_KEY) - SHA-256 checksum verification to detect tampering or corruption
- Dual-mode PostgreSQL backup: Direct mode (network) or Docker mode (container exec)
- Scheduled backups with cron-style scheduling (internal or external triggers)
- Smart retention policies: Keep last N daily/weekly/monthly backups
- Database restore wizard with pre-restore backup and verification
- Webhook notifications with HMAC-SHA256 signed payloads
- Multi-instance support: Environment-based subdirectories for shared storage
Tech Stack
| Layer | Technology |
| ---------------- | --------------------------------------------------------------- |
| Frontend | React 19, TypeScript, Vite, Tailwind CSS, Shadcn/ui |
| State | Redux Toolkit, TanStack Query |
| Backend | Node.js 22, Fastify, tRPC 11, Socket.io |
| Database | PostgreSQL 15, Prisma ORM |
| Graph (optional) | Kanbu Graphiti (Python/FastAPI), FalkorDB — semantic search/RAG |
| Agent (optional) | OpenClaw gateway, @kanbu/openclaw-bridge |
| Monorepo | pnpm workspaces, Turborepo |
Quick Start
Which branch? Clone
mainfor the latest stable version. Thedevelopbranch contains unreleased features and may be unstable.
Install Options
Choose your setup based on which features you need:
| Option | Command | Includes |
| ------------------------------- | -------------------------------------------------------------- | ------------------------------------------------------------------------------ |
| Basic | docker compose -f docker/docker-compose.selfhosted.yml up -d | PostgreSQL, API, Web UI |
| Full (with knowledge graph) | docker compose -f docker/docker-compose.full.yml up -d | Everything above + FalkorDB, Graphiti, semantic search, D3 graph visualization |
Full install requires an OpenAI API key for knowledge graph embe
Truncated for display — read the full file on GitHub.
Related Skills
taste-skill
108.6kAgent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
Agent-Reach
108.0kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
78.7kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
CowAgent
47.8kOpen-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
