leaper-agent 2.0.1

自进化 AI 员工框架 — 越用越聪明

🚀 Leaper Agent

Self-Evolving AI Agent Framework — Multi-Provider, Multi-Agent, with Built-in Memory

自进化 AI Agent 框架 — 多模型、多智能体、内置记忆引擎

Website · Quick Start · Features · 中国版

---

✨ Why Leaper Agent?

Most agent frameworks treat every conversation as a blank slate. Leaper Agent is different — it learns from every interaction and gets smarter over time, powered by DeepBrain's 6-layer self-evolving memory.

Built on top of Hermes Agent, enhanced with persistent memory, multi-agent orchestration, and production-ready integrations.

🎯 Features

• 🧠 Self-Evolving Memory — DeepBrain 6-layer memory: your agent remembers, consolidates, and evolves
• 🔌 Multi-Provider — OpenAI, Anthropic Claude, Google Gemini, and more. Switch with one line
• 🤝 Multi-Agent — Orchestrate multiple agents with different roles and specialties
• 🤖 Telegram Integration — Deploy as a Telegram bot in minutes
• 🎭 Role Templates — Pre-built personas (CTO, Analyst, Writer…) or create your own
• 🔧 MCP Support — Model Context Protocol for tool/plugin extensibility
• 📡 Streaming — Real-time streaming responses out of the box

🚀 Quick Start

# 1. Install
pip install leaper-agent

# 2. Set your API key (pick one)
export OPENAI_API_KEY=sk-xxx
# or: export ANTHROPIC_API_KEY=sk-ant-xxx
# or: export GEMINI_API_KEY=xxx

# 3. Initialize
leaper-agent init

# 4. Chat!
leaper-agent chat

As a Telegram Bot

export TELEGRAM_BOT_TOKEN=your-token
leaper-agent serve --telegram

Multi-Agent Setup

from leaper_agent import Agent, Team

researcher = Agent(role="researcher", provider="openai", model="gpt-4o")
writer = Agent(role="writer", provider="anthropic", model="claude-sonnet-4-20250514")

team = Team(agents=[researcher, writer])
result = team.run("Research and write an article about quantum computing")

🏗️ Architecture

┌─────────────────────────────────────────┐
│           Leaper Agent Core             │
├─────────┬──────────┬───────────────────┤
│ OpenAI  │ Claude   │ Gemini  │ ...     │
├─────────┴──────────┴───────────────────┤
│         Role Templates & Tools          │
├─────────────────────────────────────────┤
│         Multi-Agent Orchestration       │
├─────────────────────────────────────────┤
│         🧠 DeepBrain Memory Engine      │
│    6-Layer Self-Evolving Knowledge      │
│    ┌──────────────────────────────┐     │
│    │  4-Gate Quality Control      │     │
│    │  Relevance → Novelty →       │     │
│    │  Consistency → Utility       │     │
│    └──────────────────────────────┘     │
└─────────────────────────────────────────┘

🔌 Supported Providers

Provider	Models	Status
OpenAI	GPT-4o, GPT-4o-mini, o1/o3	✅
Anthropic	Claude Sonnet, Opus, Haiku	✅
Google	Gemini 2.0, 2.5	✅
Ollama	Any local model	✅

🎭 Built-in Role Templates

# Use a pre-built role
leaper-agent chat --role cto
leaper-agent chat --role analyst
leaper-agent chat --role writer

# Create your own
leaper-agent role create my-role

🧠 Memory in Action

from leaper_agent import Agent

agent = Agent(provider="openai", model="gpt-4o", memory=True)

# Session 1
agent.chat("My name is Ray, I'm building an AI startup")

# Session 2 — agent remembers!
agent.chat("What's my name?")  # → "Your name is Ray..."

🇨🇳 China Version / 中国版

For users in China with Chinese LLM providers (通义千问、DeepSeek、文心一言):

pip install leaper-agent-cn

→ Leaper Agent CN

📄 License

BSL-1.1 — see LICENSE for details.

🤝 Contributing

We welcome contributions! See CONTRIBUTING.md.

📧 Contact: [e-mail]

---

Built with ❤️ by Deepleaper Technology / 跃盟科技

Agents that learn. Agents that evolve. Agents that remember.