Axe: A 12MB CLI for Single-Purpose LLM Agents

What Axe Is

Axe is a 12MB Go binary with two dependencies (cobra, toml) that replaces AI frameworks with a Unix-inspired approach to LLM agents. Instead of long-lived chatbot sessions, it runs single-purpose agents defined in TOML configuration files. Each agent has a focused job like code reviewing, log analysis, or commit message writing.

Core Features

• TOML-based configuration: Declarative, version-controllable agent definitions with system prompts, model selection, skill files, and context files
• Stdin piping: git diff | axe run reviewer works directly
• Sub-agent delegation: Agents can call other agents via LLM tool use with depth limiting and parallel execution
• Persistent memory: Timestamped markdown logs carry context across runs with LLM-assisted garbage collection
• Multi-provider support: Works with Anthropic, OpenAI, Ollama (local models), or any models.dev format
• Built-in tools: Web search, URL fetch, and path-sandboxed file operations (read, write, edit, list) locked to agent's working directory
• MCP support: Can connect any MCP server to agents
• Skill system: Reusable instruction sets shared across agents
• JSON output: Structured output with metadata for scripting
• Dry-run mode: Inspect resolved context without calling the LLM

Installation & Setup

Requires Go 1.24+. Install via:

go install github.com/jrswab/axe@latest

Or build from source:

git clone https://github.com/jrswab/axe.git
cd axe
go build .

Initialize configuration:

axe config init

Creates directory structure at $XDG_CONFIG_HOME/axe/ with sample skill and default config.toml for provider credentials.

Usage Examples

Create and run an agent:

axe agents init my-agent
axe agents edit my-agent
axe run my-agent

Pipe data from other tools:

git diff --cached | axe run pr-reviewer
cat error.log | axe run log-analyzer

Copy example agents from the examples/ directory:

cp examples/code-reviewer/code-reviewer.toml "$(axe config path)/agents/"
cp -r examples/code-reviewer/skills/ "$(axe config path)/skills/"
export ANTHROPIC_API_KEY="your-key-here"
git diff | axe run code-reviewer

Docker Deployment

Build the image:

docker build -t axe .

Multi-architecture builds (linux/amd64, linux/arm64) via buildx:

docker buildx build --platform linux/amd64,linux/arm64 -t axe:latest .

Run an agent with mounted config:

docker run --rm \
  -v ./my-config:/home/axe/.config/axe \
  -e ANTHROPIC_API_KEY \
  axe run my-agent

Pipe stdin with -i flag:

git diff | docker run --rm -i \
  -v ./my-config:/home/axe/.config/axe \
  -e ANTHROPIC_API_KEY \
  axe run my-agent

Who It's For

Developers who want to automate specific AI tasks without framework overhead, especially those already using Unix tools, git hooks, cron, or CI pipelines.