A developer framework that provides a unified multi-provider LLM API and an interactive coding agent CLI with tool calling. It is self-hosted and locally runnable, but core agent functionality requires external LLM provider API credentials, and it does not include a built-in permission system for sandboxing.
Project overview
Consolidates provider-specific LLM APIs, agent state management, and an interactive terminal interface into a single resource.
Project type
AI Agent · AI Coding · Infrastructure
Use cases
Coding & Development · Automation
Deployment
Refer to project documentation
License
MIT
Best for
Developers needing an agent runtime and interactive terminal UI for coding tasks who can provide their own external LLM API credentials.
Key capabilities
An interactive coding agent that runs in the terminal with tool calling capabilities.
Provides a unified API to interact with LLMs from OpenAI, Anthropic, Google, etc.
Limitations and risks
Does not include a built-in permission system for restricting filesystem, process, network, or credential access.
By default, it runs with the permissions of the user and process that launched it.
Getting started
Setup involves installing the package via npm or running from source and configuring external LLM provider API credentials before running the CLI.
README: Pi does not include a built-in permission system for restricting filesystem, process, network, or credential access. By default, it runs with the permissions of the user and proce…
README: - **Plain Docker**: run the whole `pi` process in a local container for simple isolation.