axons 0.1.1

AI-powered coding agent — Claude Code-style CLI for AI-assisted coding

Axon — AI-powered coding agent

Claude Code-style CLI for AI-assisted code generation, editing, and debugging.

Installation

pip install axon

Quick start

# Set your API key
axon key sk-your-api-key

# (Optional) Set a custom model
axon model gpt-4o

# Or use a local model via Ollama
axon url http://localhost:11434/v1

# Start working
axon

Usage

axon                          Start interactive session
axon key <api_key>            Set your API key
axon model <model_name>       Set the LLM model
axon url <base_url>           Set custom API base URL
axon version                  Show version

Interactive session commands

Command	Description
Just type a task	The agent works on it with full tool access
/btw <question>	Quick side question (no agent context)
/model <name>	Switch model mid-session
/key <key>	Update API key
/url <url>	Update base URL
/workspace <dir>	Change workspace
/clear	Reset conversation
/exit	Quit

Environment variables

• OPENAI_API_KEY — OpenAI API key
• ANTHROPIC_API_KEY — Anthropic API key
• AXON_API_KEY — Custom provider API key
• OPENAI_BASE_URL — Custom API base URL
• AXON_MODEL — Default model name
• AXON_PROVIDER — Provider (openai, anthropic, ollama, custom)

Features

• Multi-provider: OpenAI, Anthropic Claude, Ollama, custom endpoints
• 8 built-in tools: read/write/edit files, run commands, search code, list files, ask questions
• Persistent context: remembers conversation across tasks
• Safety: dangerous command detection, output size limits, silent-loop protection
• MCP server: can be used as an MCP tool server

How it works

1. Run axon to start an interactive session
2. Type a coding task — the agent plans and executes it step by step
3. The agent reads/writes files, runs commands, searches code — all through your terminal
4. Multiple tasks maintain conversation context automatically