A self-hosted personal AI assistant in Rust with WASM-sandboxed tools, encrypted credential storage, and a web UI, inspired by OpenClaw.
ClawRunr/JavaClaw
A Java-based personal AI assistant on your own devices, acting as a gateway for an LLM agent across chat channels, tasks, shell and web.
- Stars
- 753
- Forks
- 167
- Open issues
- 27
- Last commit
- 15 days ago
What is it?
JavaClaw is a personal AI assistant written in Java, a Java port of the OpenClaw idea. It serves as a control plane for an AI agent that receives messages from several channels, schedules and tracks tasks stored as Markdown files, runs shell commands, reads and writes local files, and browses the web. It is built on Spring Boot, Spring AI and JobRunr, and background work runs through JobRunr with a dashboard on port 8081.
The agent can use OpenAI, Anthropic or a local Ollama model, and it connects to external tool servers through the Model Context Protocol. Skills are added by dropping a SKILL.md file into a workspace folder or by packaging them as jars on the classpath. A guided onboarding page at /onboarding handles provider, credentials, channels and plugins.
The project began as a demo of JobRunr and is now positioned as an open invitation for the Java community to build Java-based AI agents together.
Key features
- 01Chat UI over WebSocket, plus Telegram, Discord and WhatsApp channels via wacli
- 02Tasks scheduled once, delayed or via cron, stored as Markdown files
- 03Skills loaded from workspace folders or from skill jars on the classpath
- 04Supports OpenAI, Anthropic and Ollama as LLM providers
- 05MCP client for connecting external tool servers
- 06Agent can read and write files and run bash commands on the local machine
Use cases
- Running a self-hosted assistant that answers chat messages on your own hardware with a local Ollama model
- Scheduling recurring automated tasks written as readable Markdown files
- Extending a Java-based agent with custom skills or MCP tool servers
- Experimenting with Spring AI and JobRunr in an agent project
Quick start
- 1
Check that Java 25 and Gradle (or the included wrapper) are installed
- 2
Start the app from the cloned repo
./gradlew :app:bootRun - 3
Open the onboarding page in your browser and complete setup
Things to know
- Requires Java 25 and an LLM API key (OpenAI or Anthropic) or a running Ollama instance.
- The agent can run shell commands on your machine, so it needs careful configuration.
- WhatsApp support relies on the unofficial WhatsApp Web protocol and is used at your own risk.
- Spring AI is pinned to a 2.0.0-SNAPSHOT dependency, which is not a stable release.
Topics
FAQ
- What is JavaClaw?
- A Java-based personal AI assistant on your own devices, acting as a gateway for an LLM agent across chat channels, tasks, shell and web. JavaClaw is a personal AI assistant written in Java, a Java port of the OpenClaw idea. It serves as a control plane for an AI agent that receives messages from several channels, schedules and tracks tasks stored as Markdown files, runs shell commands, reads and writes local files, and browses the web. It is built on Spring Boot, Spring AI and JobRunr, and background work runs through JobRunr with a dashboard on port 8081.
- Is JavaClaw free to use? What license does it have?
- JavaClaw is open source under the LGPL-3.0 license. Check the license terms before using it in commercial products.
- Is JavaClaw still maintained?
- Actively maintained. The last commit to JavaClaw was on Oct 7, 2026.
- What can I use JavaClaw for?
- Running a self-hosted assistant that answers chat messages on your own hardware with a local Ollama model Scheduling recurring automated tasks written as readable Markdown files Extending a Java-based agent with custom skills or MCP tool servers Experimenting with Spring AI and JobRunr in an agent project
- How do I install JavaClaw?
- Check that Java 25 and Gradle (or the included wrapper) are installed Start the app from the cloned repo: ./gradlew :app:bootRun Open the onboarding page in your browser and complete setup
- Where can I get JavaClaw?
- The source code and installation instructions are on GitHub: https://github.com/ClawRunr/JavaClaw
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Self-hosted personal AI assistant with bio-inspired memory that is fused, linked and forgotten over time, exposed over MCP and compatible with OpenClaw skills.
Open-source AI agent application platform built on Java, LangChain4j, Spring Boot and Vue 3, covering models, agents, RAG, Skills, MCP and text-to-SQL analysis.
Self-hosted Java/Spring Boot AI assistant platform with multi-user workspaces, multi-provider model failover, an LLM Wiki, MCP and Skills, and IM channel integrations.
Self-hosted, local-first personal AI agent with layered memory, tool permissions, workflows and IM channel integration.
A self-improving AI agent that creates and refines skills from experience, runs on many model providers, and works through CLI and messaging apps.