cliver 0.2.0

Personal AI Lab

CLIver

Personal AI Lab — Experimenting with AI agents in your terminal, as a Python library, or as a gateway service.

---

CLIver is an AI Agent research and experimentation project. It provides the building blocks to explore agent architectures — Re-Act loops, tool calling, skill activation, memory systems, and multi-agent orchestration — through a terminal CLI and an admin web UI.

A lab, not a product. CLIver is designed for tinkering. Swap models mid-conversation. Test how different system prompts change behavior. Add custom skills and tools to see what sticks.

Full-stack agent experimentation. Run interactively in the terminal, embed AgentCore as a Python library in your own experiments, or deploy as a gateway service with cron scheduling and messaging platform adapters.

Quick Start

pip install cliver

# Configure your LLM provider
cliver provider add --name deepseek --type openai \
  --api-url https://api.deepseek.com --api-key "DEEPSEEK_API_KEY"

# Configure a model
cliver model add -n deepseek-v4-flash --provider deepseek

# Start experimenting
cliver "What can you help me with?"

Built-in Capabilities

• Model-agnostic — OpenAI + Anthropic protocols, any provider, switch on the fly
• Built-in tools — file I/O, shell, web search, browser, memory, todo tracking, image generation
• Skills system — LLM-activated domain expertise following the Agent Skills spec
• Permission control — default, auto-edit, and YOLO modes with per-tool overrides
• Persistent memory — remembers preferences and context across sessions
• Gateway mode — daemon with cron scheduling, admin web UI, and messaging platform adapters
• Embeddable API — AgentCore.chat() and .stream() for your own Python experiments

Documentation

Full documentation at cliver-project.github.io/CLIver

Development

git clone https://github.com/cliver-project/CLIver.git
cd CLIver

make init     # Set up dev environment
make test     # Run tests
make lint     # Lint and format check
make format   # Auto-fix lint/format issues

make docs-serve  # Start docs website (hot reload)
make docs-build  # Build docs to static files

License

Apache 2.0