CLIver Documentation¶
CLIver is a general-purpose AI agent for your terminal. It is model-agnostic, extensible, and safe by design.
Getting Started¶
-
:material-download: Installation
Install CLIver via pip, Docker, or from source.
-
:material-cog: Configuration
Set up LLM providers, MCP servers, and permissions.
-
:material-chat: Chat Usage
Learn interactive and batch chat modes.
-
:material-lightning-bolt: Skills
Teach CLIver new domains with skill files.
Core Guides¶
| Guide | What You'll Learn |
|---|---|
| Memory & Identity | Persistent knowledge and agent profiles |
| Permissions | Control what tools CLIver can execute |
| Workflows | Multi-step task orchestration with LangGraph |
| Session Management | Conversation history and compression |
| Gateway | Deploy with Telegram, Discord, Slack, Feishu |
Extend & Integrate¶
| Guide | What You'll Learn |
|---|---|
| Extensibility & API | Use AgentCore as a Python library |
| Roadmap | Planned features and how to contribute |
Quick Example¶
# Install and start
pip install cliver
cliver
# Add a model
cliver model add --name deepseek --provider openai --url https://api.deepseek.com
# Add an MCP server
cliver mcp add --name filesystem --transport stdio --command uvx -- mcp-server-filesystem
# Chat with tools
cliver "List all Python files in this directory and summarize the largest one"