🚀 Quick Start · Chapter 03

Configure Your First AI Agent in 3 Steps

Start from scratch, create your own exclusive AI assistant in 3 steps. The entire process takes only 5 minutes.

🎯
Simplified 3-Step Core
Set EasyClaw's default main Agent as "Main Agent" → Configure your digital employees → Save and run. Subsequent sub-Agents and cross-channel collaboration can be iterated gradually.

Step 1: Tell the Main Agent "Who You Are"

If you have multiple Agents, it's recommended to explicitly set the default main Agent as the "Main Agent" role. In the dialogue box, directly enter the rules you want to set, then ask the Agent to save important content to MEMORY.md:

Step 1 illustration
💡
Why write to MEMORY.md?
If you just tell the Agent verbally, it will forget in the next conversation. Information written to MEMORY.md will be retained permanently, and the Agent will prioritize reading it at every startup.

Step 2: Configure Your Digital Employees

In the "Digital Employees" panel, fill in the following information as prompted to quickly create a dedicated Agent:

Step 2 illustration
1

Fill in Basic Information

Enter the Agent's name and responsibility description. For example: "Daily News Assistant", "Code Review Specialist", "Customer Service Bot". The more specific the description, the more precise the Agent's execution.

2

Associate Skills (Optional)

Specify the Skills this Agent needs. For example, a customer service Agent can be associated with "Knowledge Base Query" and "Lark Message Sending" Skills.

3

Save and Test

After saving, find the newly created Agent in the "Digital Employees" list, click to enter conversation, send a test command to verify effectiveness.

⚠️
Not Recommended to Frequently Switch Between Main and Sub-Agents
Each Agent's context memory is mutually independent. If you need to pass core conclusions from one conversation to another Agent, you need to explicitly write them to MEMORY.md, not just directly switch.

Step 3: Store Configuration and Run

After configuration, you'll see your newly created Agent in the "Digital Employees" list. Click the corresponding Agent to start conversing with it in the interface, or bind it to a phone remote channel to initiate tasks anytime, anywhere.

Step 3 illustration

Quick Start Common Questions

How do multiple Agents collaborate?

  • Multiple Agents can mutually know each other's existence (share information through MEMORY.md)
  • Each Agent's conversation memory is mutually independent, but can share the same MEMORY.md
  • One Agent can directly call other Agents to execute sub-tasks—this is precisely multi-Agent collaboration mode

How to tell the Agent about yourself for the first time?

Just state it directly in the conversation, then ask the Agent to write key content to memory files. Next time you tell the same Agent, just reference it directly without needing to repeat information.