Pi.dev

Overview

Terminal-based AI coding assistant for interactive development workflows.

PI.dev is a minimal terminal-based coding agent harness for developers, teams, and automation workflows.It offers interactive TUI, print/JSON, RPC, and SDK modes for local use, scripting, inter-process integrations, and embedding.

Extensions support TypeScript modules, skills, prompt templates, themes, and packages that install via npm or git and can be shared as pi-packages.Authentication supports API keys and OAuth across 15+ providers and hundreds of models (Anthropic, OpenAI, Google, Azure, Bedrock, Mistral, Hugging Face, and more) with mid-session model switching and custom providers via models.

json.Session trees, auto-compaction summaries, dynamic context injection, and RAG/long-term memory primitives help maintain usable context for large conversations and codebases.Skills and extensions expose tools, commands, keyboard shortcuts, events, and TUI components for sub-agents, plan mode, permission gates, SSH/sandbox execution, and custom editors.

CLI features include package install/update/list/configure, version pinning, test mode for repos, queuing and steering messages, and export/share options (HTML, GitHub Gist) for reproducible agent workflows.

Use Cases

• Build an interactive terminal-based coding assistant for your engineering team using pi.dev's TUI and dynamic model switching to run code, debug sessions, manage session/context, and reproduce paired-programming workflows without leaving the terminal.
• Automate code generation, refactors, CI checks and deployment tasks with pi.dev's CLI tooling and RPC/SDK integration, combining multi-provider model authentication, TypeScript extension modules, and reproducible agent workflows to enforce standards and accelerate delivery.
• Create a project-aware developer platform that leverages pi.dev's RAG and long-term memory primitives plus TypeScript packages to surface project-specific knowledge, searchable context, and persistent memories for onboarding, incident response, and maintainable documentation.

Who Is It For

• Software developers
• Automation engineers
• Machine learning engineers
• Prompt engineers
• Tool authors