agent-creator

What This Skill Does

The agent-creator skill acts as the primary administrative engine for the OpenClaw ecosystem. It automates the complex, multi-step process of bootstrapping a new AI agent within the framework. Rather than manually creating directories, copying templates, and editing core configuration files like openclaw.json, this skill orchestrates the entire workflow. It ensures that every new agent is instantiated with the mandatory identity and behavioral frameworks (IDENTITY.md, SOUL.md, USER.md, AGENTS.md, and MEMORY.md), preventing "empty agent" errors that occur when template files are missing. By standardizing the creation process, it guarantees that all agents are immediately recognizable by the OpenClaw core, while providing a seamless bridge for optional integrations like Feishu messaging channels.

Installation

To integrate this administrative tool into your environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/axelhu/openclaw-agent-creator Once installed, the skill automatically hooks into the OpenClaw command processor, allowing you to invoke agent creation routines through standard natural language prompts.

Use Cases

• Rapid Prototyping: Instantly spin up a new agent for model testing or specialized task handling.
• Environmental Expansion: Add new specialized robotic or logical agents to an existing swarm.
• Channel Integration: Standardize the configuration process for linking internal agents to external communication platforms like Feishu.
• Identity Management: Deploy agents with specific "souls" and "memory" constraints using pre-validated project templates.

Example Prompts

1. "Create a new agent named 'ResearchBot' with a formal and analytical personality."
2. "I need to spin up a new test environment for the 'CodeReviewer' agent, please set it up with the default memory and soul profiles."
3. "Setup a new Feishu-integrated agent called 'SupportDroid' and guide me through the configuration of its API keys."

Tips & Limitations

• Strict File Integrity: Always ensure the template directory is intact. Missing files in ~/.openclaw/templates/ will result in broken agent instances.
• Verification: Always run openclaw status after running the creation skill to ensure the new agent is correctly registered in the system registry.
• Configuration Caution: When manually editing openclaw.json after the initial creation, ensure all JSON syntax is valid; a single malformed character can cause the entire OpenClaw agent pool to crash on load.
• Manual Auth: While this skill handles directory structure, sensitive credentials like API keys for models should still be treated with high security and should not be stored in plain text if possible.