Skip to content
Projects

Puteitous/HippoBuddy has 56 GitHub stars and 11 forks. GitHot measured 0 stars in 24 hours, +1 in 7 days, and +1 in 15 days. Last checked 2026-11-15 20:35 UTC. Last push 2026-10-07.

How these counts are measured

Measured star growth

in 24h
0
in 7 days
+1
in 15 days
+1
in 1 month
Update delayed

Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.

About

HippoBuddy — AI-powered desktop assistant for chat, coding, and office workflows.

Productivity

docs/README.en.md

Original

HippoBuddy HippoBuddy

AI-powered desktop assistant for chat, coding, and office productivity.

简体中文 | English

Java 21 Electron Release Stars License Platform Last Commit

HippoBuddy 主界面


Download

Platform Download
Windows HippoBuddy Setup
macOS (Intel) HippoBuddy.dmg
macOS (Apple Silicon) HippoBuddy-arm64.dmg
Linux (AppImage) HippoBuddy.AppImage

📖 Online documentation: https://www.hippobuddy.cn/

🪞 Gitee mirror repository: https://gitee.com/putetou/HippoBuddy

🔗 Baidu Netdisk: https://pan.baidu.com/s/1L78e0I7N4zaz_yZsVTeW7A?pwd=pfga

💬 QQ Group (1102524202) — feedback, feature requests, help, and casual chat


Features

Feature Description
Smart Chat Chat / Code / Office modes, switch anytime
AI Coding Understand project context, generate & refactor code
File Ops Read, write, edit, delete with diff preview & rollback
Sessions Create, rename, delete, fork discussions
Office Read/Write Native parse & generation for Word / Excel / PPT / CSV
Subagent Fork, cancel, and run parallel subagents (worker agents)
Memory Extract, consolidate & retrieve cross-session memory
MCP Extension Built-in SSE & Stdio MCP clients, plug in external tools
Skill System Load and orchestrate reusable skill libraries on demand
Toolbox Token stats, terminal, browser, live monitor
Onboarding Spotlight tour on first launch

Chat 与预览面板
Chat panel and preview panel working together


Why HippoBuddy?

Compared to other AI agent tools (Codex, Claude Code, Copilot, Kimi, Trae Work, WorkBuddy):

Dimension HippoBuddy
Open Source Full source, Apache 2.0 license
Zero Setup No login, no accounts, no third-party services — download and use
LLM Visibility Every tool call and thinking step is visible in real-time
Code Editing Full read/write/edit with diff preview and rollback
Office Documents Built-in viewer for PDF, Word, Excel, PPT, and more
File Change System Track changes at file and session level, rollback anytime
Context & Token Monitor Real-time token stats, context usage, LLM monitoring
Built-in Tools 16+ built-in tools: terminal, browser, search, code analysis, etc., extensible via MCP
Performance Lightweight desktop app, Java virtual-thread concurrency
UI Design Minimalist and clean
Platform Desktop (Windows / macOS / Linux)

Roadmap

HippoBuddy is under active iteration. Core capabilities (MCP, subagents, memory, Office read/write, skill system) are already shipped and will be continuously polished:

  • Requires personal LLM API key + web search tool config
  • Subagent, MCP, and Memory are supported but being hardened for stability & ease of use
  • Automated task pipeline and browser automation are planned
  • More complex Office layout editing will be improved further
  • Third-party software integration is still limited
  • Designed for personal task efficiency, not 24/7 online service

🎬 Video Intro

A 6-minute quick look at HippoBuddy:

HippoBuddy Intro Video
Watch Intro Video on Bilibili


Quick Start

Option 1: Desktop (Recommended)

Download installer -> Install -> Launch -> Start using

Option 2: From Source

# 1. Build the frontend (Vite output goes to src/main/resources/static)
cd frontend && npm install && npm run build && cd ..

# 2. Compile and package the Java backend (includes the static assets above)
mvn package -DskipTests

# 3a. Launch desktop (Electron)
cd electron && npm install && npm start

# 3b. Or run web-only (no Electron)
mvn exec:java -Dexec.mainClass="com.example.agent.WebApplication"

Configuration — When running from source, the app auto-creates config.yaml from config.yaml.example on first launch. Edit it with your LLM settings:

llm:
  api_key: ${DEEPSEEK_API_KEY:-your-api-key-here}
  model: deepseek-v4-flash
  base_url: https://api.deepseek.com

Supports DeepSeek / Claude / GPT / Ollama. See config.yaml.example for full reference.


Tech Stack

Layer Technology
Desktop Shell Electron 35
Frontend React 18 + TypeScript + Vite 5
State Management Zustand 4
Code Editor CodeMirror 6
Backend Java 21 + Virtual Threads
Build Maven 3.9 + npm
AI Protocol OpenAI / Claude / Ollama
Testing JUnit 5 + Vitest + Testing Library

Project Structure

src/main/java/com/example/agent/
├── WebApplication.java           Web entry
├── DesktopApplication.java       Desktop entry
├── core/                         DI, event bus, security blockers
├── llm/                          LLM clients (OpenAI, Claude, Ollama...)
├── tools/                        Built-in tools (16, MCP-extensible)
├── execute/                      Agent conversation loop
├── subagent/                     Multi-agent system
├── mcp/                          MCP protocol
├── memory/                       Long-term memory
├── context/                      Context budget & compaction
├── session/                      Session storage & transcripts
├── web/                          HTTP handlers & SSE streaming
│   └── orchestrator/             Task orchestration (DAG)
├── application/                  Conversation application service
├── service/                      Token estimation, title generation
├── desktop/                      Desktop workspace context
├── console/                      Console interaction
├── progress/                     Progress & diff preview
├── logging/                      Logging & metrics collection
├── prompt/                       Prompt library & management
├── domain/                       Rules, skills, content truncation
└── config/                       Configuration models

Documentation

Title Link
HippoBuddy — Introduction docs/intro
Quick Start docs/quick-start
HippoBuddy Architecture Philosophy docs/architecture/philosophy
AI Agent Usage Tips docs/guides/agent-mindset
What Is an AI Desktop App Loading on Startup? docs/guides/startup-loading

License

Apache License 2.0