What is ZhikunCode?

ZhikunCode is a self-hosted AI coding platform that provides a complete browser-based development environment. It features a three-tier architecture with Java Spring Boot backend, React TypeScript frontend, and Python FastAPI analysis service. Users can delegate tasks to multiple specialized agents that run in parallel, with background task support and real-time activity tracking. The platform includes 40+ built-in tools for file operations, terminal, search, and browser automation, plus MCP protocol support for dynamic tool extension. Security is built-in with an 8-layer Bash sandbox, unified authorization gateway, and tool execution gateway. It natively connects to Chinese LLM providers (Qwen/DashScope, DeepSeek, Moonshot/Kimi, Zhipu GLM, MiniMax) without requiring external network access. Deployment is via Docker Compose with persistent SQLite storage for sessions and memories. The project is MIT licensed and actively developed.

Should you self-host ZhikunCode?

ZhikunCode scores 51/100 on selfhost.directory (how scoring works) and ranks #33 of 120 in AI Assistant. It has 532 GitHub stars.

It shipped 1 stable release in the past year. The latest, v1.0.0, came out on 21 April 2026.

Setup is moderate: expect some configuration beyond a single command. It is MIT-licensed and written mostly in Java.

Its closest alternative on selfhost.directory is openhands (87/100, 92 086 stars). It scores higher overall, so compare the two before you pick. See ZhikunCode vs openhands side by side.

Features

8-layer security sandbox
Background tasks
Browser-based Web UI
Cross-session memory
MCP protocol support
Multi-Agent Collaboration
Multimodal image chat
Native Chinese LLM support
Plugin System
Project and session trust boundaries
Runtime verification framework
Skill System
Smart context management
Voice interaction (ASR/TTS)

FAQ

What if port 8080 is already in use?

Change ZHIKUN_PORT in .env (e.g., ZHIKUN_PORT=9090), then docker compose down && docker compose up -d.

How to troubleshoot Docker deployment issues?

Check docker ps -a, docker logs zhikuncode, docker inspect health. Common issues: missing API key, incomplete image build (run docker compose up --build), insufficient memory (adjust deploy.resources.limits.memory in docker-compose.yml).

Why Java + React + Python three-tier architecture?

Java 21/Spring Boot for strong typing, enterprise ecosystem, WebSocket, virtual threads; React 18/TypeScript for componentization, type safety, Vite, TailwindCSS; Python FastAPI for mature code analysis/AST ecosystem, async performance, isolation from main backend.

How to contribute code?

Fork repo, create feature branch, commit, create PR. See CONTRIBUTING.md. Dev environment needs JDK 21, Node.js 22+, Python 3.11-3.12, Maven 3.9+.

Does it conflict with VS Code extensions like Copilot or Cline?

No conflict. ZhikunCode is an independent web app that doesn't depend on any IDE or require plugins.

Does it support offline/air-gapped deployment?

Yes, fully offline with Ollama for local models, or use Chinese LLMs with domestic network access only.

Where is data stored and is it secure?

Session data in SQLite database in Docker volume zhikun-data, project code via volume mount, API keys only in .env and container env vars. No telemetry runs.

What are the Docker deployment requirements?

Docker 20.10+, Docker Compose V2, 4GB+ RAM, and network access to LLM API endpoints.

Which LLMs are supported?

All OpenAI API-compatible models including Qwen/DashScope, DeepSeek, Moonshot/Kimi, Zhipu GLM, MiniMax, OpenAI, Anthropic Claude, and local models via Ollama/vLLM.