afm-cli

AFM CLI

A command-line interface for running Agent-Flavored Markdown (AFM) agents.

This package provides everything you need to run AFM agents, including:

• afm-core: Parser, validation, and interface implementations
• afm-langchain: LangChain-based execution backend with support for OpenAI and Anthropic

Installation

pip install afm-cli

Quick Start

1. Set your API key

export OPENAI_API_KEY="your-api-key-here"
# or
export ANTHROPIC_API_KEY="your-api-key-here"

2. Create an AFM file

Create a file named agent.afm.md:

---
name: Friendly Assistant
interfaces:
  - type: consolechat
model:
  provider: openai
  name: gpt-4o

---
# Role
You are a helpful and friendly AI assistant.

# Instructions
- Be concise but thorough in your responses
- If you don't know something, say so honestly
- Always be polite and professional

3. Run the agent

afm run agent.afm.md

This will start an interactive console chat with your agent.

Usage

afm run <file>

Run an AFM agent file and start its configured interfaces.

# Run with default settings
afm run agent.afm.md

# Run on a specific port (for web interfaces)
afm run agent.afm.md --port 8080

afm validate <file>

Validate an AFM file without running it.

afm validate agent.afm.md

afm framework list

List available execution backends.

afm framework list

Configuration

Environment Variables

• OPENAI_API_KEY - Required for OpenAI models
• ANTHROPIC_API_KEY - Required for Anthropic models

CLI Options

• -p, --port PORT - Port for web interfaces (default: 8085)
• --help - Show help message

Features

Interface Types

AFM agents can expose multiple interfaces simultaneously:

• consolechat: Interactive terminal-based chat (default)
• webchat: HTTP API with built-in web UI
• webhook: Webhook endpoint for event-driven agents

MCP Tool Support

Agents can use external tools via Model Context Protocol (MCP).

For more examples and detailed documentation, see AFM Examples.

Docker

You can also run agents using Docker:

# Build the image
docker build -t afm-langchain-interpreter .

# Run with an AFM file mounted
docker run -v $(pwd)/agent.afm.md:/app/agent.afm.md \
  -e OPENAI_API_KEY=$OPENAI_API_KEY \
  -p 8085:8085 \
  afm-langchain-interpreter afm /app/agent.afm.md

Documentation

For more detailed documentation, see the project README.

License

Apache-2.0