Project Configuration (aigne.yaml)

The aigne.yaml file is the heart of your AIGNE project. It serves as the central manifest, defining the project's metadata, language model settings, and the relationships between its core components like agents and skills. A well-structured aigne.yaml is the first step toward building a powerful and organized agent.

This file uses the YAML format, which is designed to be human-readable and easy to write.

Core Configuration Keys#

Let's break down the primary sections you'll find in a typical aigne.yaml file.

Project Metadata#

These fields provide basic information about your project.

name: test_aigne_project
description: A test project for the aigne agent

Chat Model (chat_model)#

This is a critical section where you configure the Large Language Model (LLM) that will power your agents. AIGNE provides a flexible way to define the model provider, name, and other parameters.

Key	Type	Description
provider	string	The LLM provider (e.g., openai). Can also be specified as a prefix in the model key.
name / model	string	The specific model to use (e.g., gpt-4o-mini).
temperature	number	A value between 0 and 2 that controls the randomness of the model's output. Higher values result in more creative responses.

Here are a couple of common ways to configure the model:

Example 1: Using provider and name

This is a clear, explicit way to define the model.

chat_model:
  provider: openai
  name: gpt-4o-mini
  temperature: 0.8

Example 2: Using a prefixed model key

This is a more concise shorthand format.

chat_model:
  model: openai:gpt-4o-mini
  temperature: 0.8

Agents (agents)#

The agents key lists all the agent definition files (.yaml) included in your project. Each file listed here defines the behavior, prompts, and tool usage for a specific agent.

# ... other configurations
agents:
  - chat.yaml

Skills (skills)#

The skills key lists the executable code or definitions that provide your agents with tools and capabilities. These can be JavaScript files (.js) containing functions or other YAML files that define complex skills.

# ... other configurations
skills:
  - sandbox.js
  - filesystem.yaml

Service and CLI Exposure#

You can also configure how your agents are exposed to the outside world, whether through a server or as command-line tools.

• mcp_server: Configures which agents are served via the Model Context Protocol (MCP) when you run the aigne serve-mcp command.
• cli: Configures which agents can be run directly from the command line.

# ... other configurations
mcp_server:
  agents:
    - chat.yaml

cli:
  agents:
    - chat.yaml

Complete Example#

Here is a complete aigne.yaml file that brings all these elements together:

name: test_aigne_project
description: A test project for the aigne agent

chat_model:
  model: openai:gpt-4o-mini
  temperature: 0.8

agents:
  - chat.yaml

skills:
  - sandbox.js

mcp_server:
  agents:
    - chat.yaml

cli:
  agents:
    - chat.yaml

With this configuration file in place, you have a solid foundation for your project. The next step is to define what your agents and skills actually do.

Now that you understand how to configure the project, let's explore the core components in more detail. Continue to the next section to learn about Agents and Skills.