Powered by AIGNE DocSmith

AIGNE CLI

Creating AI Agents

Using AIGNE's YAML configuration approach, you can quickly create powerful AI agents. Through simple configuration files, you can define AI behavior, input/output formats, and skills.

Basic Structure#

name: chat
description: Chat agent
instructions: |
You are a helpful assistant that can answer questions and provide information on a wide range of topics.
Your goal is to assist users in finding the information they need and to engage in friendly conversation.
input_schema:
type: object
properties:
message:
type: string
description: User message
required:
- "message"
output_schema:
type: object
properties:
response:
type: string
description: AI response
required:
- "response"
output_key: text
skills:
- plus.js

Configuration Options#

  • name: Unique identifier for the agent
  • description: Brief description of the agent's functionality and purpose
  • instructions: Detailed instructions guiding the agent's behavior (using YAML's multi-line text format)
  • input_schema: [Optional] JSON Schema definition for input parameters
    • type: Input data type (top level must be object)
    • properties: Detailed definition of input parameters
    • required: List of parameters that must be provided
  • output_schema: [Optional] JSON Schema definition for output results (only use when structured data output is needed)
    • type: Output data type (top level must be object)
    • properties: Detailed definition of output results
    • required: List of parameters that must be returned
  • output_key: [Optional] Key name for output text (defaults to $message, only effective when there's no output_schema)
  • skills: [Optional] List of tools the agent can use (JavaScript files implementing specific functions)
  • memory: [Optional] Enable the agent's conversation memory feature. Can be:
    • Boolean value (true to enable, false to disable)
    • Object containing configuration options:
      • subscribe_topic: Array of memory topics the agent should subscribe to


Through the above configuration, you can create feature-rich AIGNE agents. Main capabilities include:

  • Create AI assistants with different behavior patterns
  • Equip AI with various skills and tools
  • Enable conversation memory and learning capabilities
  • Integrate with other systems and resources

Recommendation: Start with a simple chatbot and gradually add more features to expand AI agent capabilities.

Agent Definitions
Creating MCP Agents