github.devotd.
NewAI

sigoden/aichat v0.30.0 — A Rust CLI that bundles multiple LLM providers

sigoden/aichat is a Rust tool that brings the model APIs of several LLM providers into one command-line interface, offering shell command conversion, RAG, tool calling and a built-in API server; some details cannot be confirmed from public materials.

Repositorysigoden/aichat ↗

sigoden/aichat is a Rust tool that brings the model APIs of several LLM providers into one command-line interface, offering shell command conversion, RAG, tool calling and a built-in API server; some details cannot be confirmed from public materials.

GitHub description: All-in-one LLM CLI tool featuring Shell Assistant, Chat-REPL, RAG, AI Tools & Agents, with access to OpenAI, Claude, Gemini, Ollama, Groq, and more.

stars
★ 10.5K
Forks
851
Language
Rust
License
Apache-2.0
Latest release
v0.30.0
Read original source ↗View on GitHub ↗

sigoden/aichat is a Rust tool that bundles the model APIs of several LLM providers — OpenAI, Claude, Gemini, Ollama, Groq and others — into a single command-line interface. It packs into one program the ability to turn natural-language requests into shell commands, an interactive REPL for giving input and receiving results, retrieval-augmented generation (RAG), and AI Tools & Agents. The latest release is v0.30.0.

Its core is choosing among several models through one interface instead of using a separate chat tool for each provider. The Shell Assistant and CMD & REPL modes handle shell command generation and conversation together, and files, directories and URLs can be used directly as input. It supports roles, sessions, macros and Function Calling, and through a built-in HTTP server it offers Chat Completions, Embeddings and Rerank APIs along with LLM Playground and LLM Arena. It is licensed under Apache-2.0, has 10504 stars, and the repository was created on March 3, 2023.

That said, what exactly changed in v0.30.0 cannot be pinned down, because the published materials contain no changelog. There are also no benchmarks measuring how natural-language requests turn into shell commands, or how accurate the tool calls are, so real-world accuracy is hard to judge from reading alone. That this tool handles several providers in one place is clear, but what the generated commands actually execute in your environment is for you to verify directly.