Repolode
Submit
mateaix

mateaix/mateclaw

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.

View on GitHubWebsite
Stars
1,170
Forks
351
Open issues
187
Last commit
2 hours ago

What is it?

MateClaw is a self-hosted AI agent platform built on Spring Boot, Spring AI and AgentScope Java. It lets a team create role-based digital employees that run chat, tool calls, plans and long-running Goals inside one deployed JAR, with RBAC, approval gates for sensitive tool use, and an audit trail.

The platform routes requests across a configured list of model providers in priority order, including DashScope, OpenAI, Anthropic, Gemini, DeepSeek, Ollama and LM Studio, and moves to the next healthy provider when one fails. Knowledge is handled by an LLM Wiki that turns uploaded documents into linked pages with citations, while workspace memory files, SKILL.md packages and MCP servers extend what each employee can do.

Users reach employees through a web console, an Electron desktop app, an embeddable webchat widget, or IM channels such as DingTalk, Feishu, WeChat Work, Telegram, Discord, QQ and Slack. Project docs describe a Docker Compose setup with PostgreSQL 16 and an H2 development mode.

Key features

  • 01Digital employees with roles, goals and per-employee tool binding
  • 02Ordered model provider chain with health tracking and cooldown failover
  • 03LLM Wiki that builds linked pages with citations from uploaded documents
  • 04MCP (stdio, SSE, Streamable HTTP) and SKILL.md skill packages for extension
  • 05Approval flows, RBAC and audit trail for sensitive tool actions
  • 06Channel integrations for DingTalk, Feishu, WeChat Work, Telegram, Discord, QQ and Slack

Use cases

  • Giving an internal team a shared AI assistant with per-user workspaces and admin controls
  • Building a knowledge base from PDFs and markdown files that answers with source citations
  • Running an AI assistant in Feishu or DingTalk groups backed by a self-hosted model
  • Delegating multi-step, long-running tasks to agents that resume after a restart

Topics

#ai-agent#java#spring-boot#multi-agent#mcp#llm#self-hosted#chatbot#agent#agent-harness

FAQ

What is mateclaw?
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. MateClaw is a self-hosted AI agent platform built on Spring Boot, Spring AI and AgentScope Java. It lets a team create role-based digital employees that run chat, tool calls, plans and long-running Goals inside one deployed JAR, with RBAC, approval gates for sensitive tool use, and an audit trail.
Is mateclaw free to use? What license does it have?
mateclaw is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is mateclaw still maintained?
Actively maintained. The last commit to mateclaw was on Oct 22, 2026.
What can I use mateclaw for?
Giving an internal team a shared AI assistant with per-user workspaces and admin controls Building a knowledge base from PDFs and markdown files that answers with source citations Running an AI assistant in Feishu or DingTalk groups backed by a self-hosted model Delegating multi-step, long-running tasks to agents that resume after a restart
Where can I get mateclaw?
The source code and installation instructions are on GitHub: https://github.com/mateaix/mateclaw

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects