AIChat — All-in-One LLM CLI with 20+ Providers

What it is

AIChat is a command-line interface that connects to 20+ LLM providers including OpenAI, Anthropic, Google, Mistral, Ollama, and more. It provides interactive chat, shell assistant mode, RAG (retrieval-augmented generation), function calling, and agent capabilities from a single binary.

AIChat targets developers and power users who prefer terminal-based workflows and want one tool for all their LLM interactions instead of separate CLIs for each provider.

How it saves time or tokens

AIChat's shell assistant mode interprets natural language commands and generates shell scripts. Instead of remembering complex command-line flags, you describe what you want and AIChat produces the command. The RAG feature indexes local files and provides context-aware answers without manual copy-pasting.

Switching between providers is a config change, so you can test the same prompt across models without leaving the terminal.

How to use

1. Install AIChat: cargo install aichat or download a prebuilt binary
2. Configure your API keys in ~/.config/aichat/config.yaml
3. Start interactive chat: aichat
4. Use shell mode: aichat -e 'find all Python files larger than 1MB'

Example

# Interactive chat with Claude
aichat --model claude-sonnet-4-20250514

# Shell assistant: describe what you need
aichat -e 'compress all PNG files in current directory to 80% quality'
# Output: find . -name '*.png' -exec convert {} -quality 80 {} \;

# RAG: index a directory and ask questions
aichat --rag-add ./docs/
aichat --rag 'What is the API rate limit?'

# Function calling with custom tools
aichat --agent 'Check the weather in San Francisco and create a calendar event'

Related on TokRepo

• Coding tools -- Terminal-based developer tools
• Automation tools -- Shell automation with AI

Common pitfalls

• Each provider requires its own API key configured in the YAML file; missing keys produce silent failures
• RAG indexing uses local embeddings by default; large document sets can be slow without GPU acceleration
• Shell assistant mode executes suggested commands only with confirmation; review commands before accepting