CLI AI Coding Agent

Ship faster with
Apollo

A terminal-native AI coding assistant that works with any OpenAI-compatible API. Edit files, run commands, and navigate codebases — all from your terminal.

apollo
You: list files in current directory
Thinking...
Executing [1/1]: ls
Thinking...
Apollo: Here's what I found...

You: add a hello function to main.go
Thinking...
Executing [1/1]: edit
Apollo: Done! Function added to main.go.

Everything you need, right in the terminal

Apollo ships with powerful built-in tools and a seamless CLI experience.

CLI Interface

Interactive terminal UI with ANSI colors, tab completion, and readline support.

Built-in Tools

ls, read, bash, and edit — all available out of the box with automatic chaining.

Markdown Rendering

Beautiful formatted output powered by Glamour — code blocks, lists, and tables rendered in your terminal.

Safe Mode

Enable --safe to review and confirm every file edit before it's applied.

Debug Mode

Inspect raw API requests with --debug to see exactly what's being sent and received.

OpenAI-Compatible

Works with any OpenAI-compatible endpoint — bring your own model and API key.

Get started in minutes

Three steps to your first conversation with Apollo.

1

Get your API key

Grab an API key from any OpenAI-compatible provider — OpenAI, Groq, Together, or your own endpoint.

2

Set the environment variable

export OPENAI_API_KEY=your_api_key

Add it to your ~/.bashrc or ~/.zshrc for persistence.

3

Run Apollo

go run . # or build and run ./apollo

Start chatting. Ask it to read files, run commands, or edit code.

Flags

-d, --debugPrint API request JSON
-s, --safeConfirm before applying edits
/usageShow token usage stats
/clearClear terminal screen
/helpShow available commands
quitExit Apollo

What's next

Upcoming features on the Apollo roadmap.

In Progress

Saving & Exporting Sessions

Save your chat sessions and export them as Markdown or HTML files. Currently in active development.

Planned

Model Picker

Switch models mid-conversation without restarting. Pick the right model for the task at hand.