A lightweight, terminal-based AI programming assistant built in Rust, pi-rs is inspired by pi-coding-agent. It targets developers and power users seeking an efficient, interactive AI tool directly within their command-line environment, offering multi-LLM support and integrated development utilities.

How It Works

This project leverages Rust for a performant, resource-efficient implementation with an interactive TUI built using ratatui. It supports numerous LLM providers (OpenAI, Anthropic, Google, Ollama, etc.) and features a robust tool system for file operations and shell commands. A flexible skill system customizes AI behavior, while context compression automatically summarizes long conversations for improved efficiency.

Quick Start & Requirements

• Installation: Clone the repository (git clone https://github.com/yourusername/pi-rs.git), navigate into the directory, and build using cargo build --release. The binary will be located at target/release/pi.
• Prerequisites: Rust toolchain, LLM provider API keys (e.g., MOONSHOT_API_KEY).
• Running: Execute ./target/release/pi --model <provider-model> "<your message>". Example: ./target/release/pi --model moonshot-v1-8k "你好，你会做什么？".
• Documentation: No explicit quick-start or demo links provided beyond the repository URL.

Highlighted Details

• Multi-Provider LLM Support: Integrates with OpenAI, Anthropic, Google, Moonshot, Ollama, Azure OpenAI, Mistral, and Groq.
• Tool System: Includes file operations (read, write, edit), shell commands (bash, grep, find, ls), and epkg (multi-source package manager for RPM, DEB, Alpine, Arch, Conda).
• Interactive TUI: User-friendly terminal interface powered by ratatui.
• Session Management: JSONL-based tree structure for conversations.
• Skill System: Load custom skills for tailored AI behavior.
• Context Compression: Automatic summarization of long dialogues.
• Sandboxing: Isolated environment with custom mounts, env vars, and types.
• Performance: Binary size ~8 MB, runtime memory ~9 MB.

Maintenance & Community

No specific details regarding maintainers, community channels (like Discord/Slack), sponsorships, or a public roadmap are present in the provided README content.

Licensing & Compatibility

• License: MIT License.
• Compatibility: Permissive MIT license allows commercial use and integration into closed-source projects.

Limitations & Caveats

The README uses placeholder URLs for cloning the repository. There is no explicit mention of alpha/beta status, known bugs, or unsupported platforms. The epkg tool's effectiveness depends on the execution environment's package sources.