CLIver is a general-purpose AI agent for your command line. It is not tied to any specific domain — with customizable system prompts, skills, workflows, and MCP integrations, you can adapt it to any task: coding, DevOps, research, writing, data analysis, or anything else you need.
CLIver is also built to be safe and controlled. A layered permission system governs every tool execution, and a structured workflow engine keeps complex tasks on track — so you get the power of autonomous AI without scattered, unpredictable behavior.
# Install via pip
pip install cliver
# Or run with Docker
docker run --rm -it --user $UID:0 -v ~/.cliver:/home/cliver/.cliver \
-e OPENAI_API_KEY ghcr.io/cliver-project/cliver
# Start chatting
cliver chat "What time is it in Beijing and London?"
# Interactive mode
cliver chat- Multi-provider LLM support — OpenAI-compatible (Qwen, DeepSeek, etc.), Ollama, vLLM
- MCP integration — Connect to any MCP server for extended tool capabilities
- Skills — LLM-driven skill activation following the Agent Skills specification
- Memory & Identity — Persistent knowledge and agent profiles across sessions
- Permissions — Layered tool permission system (default, auto-edit, yolo modes)
- Workflows — Multi-step workflow engine with pause/resume support
- Embeddable —
AgentCoreAPI can be used as a Python library
Full documentation is available at the docs site:
make init # Set up dev environment
make test # Run tests
make lint # Lint + format check
make format # Auto-fix lint and formattingSee LICENSE for details.
