The Add Agent skill automates the deployment of new, isolated agents with dedicated Telegram bots and synchronized resources.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install add-agent
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install add-agent using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Add Agent?
The Add Agent skill is a specialized automation utility designed to expand your AI capabilities by programmatically creating new agent instances. By leveraging the framework of Openclaw Skills, this tool handles the complex orchestration of directory structures, configuration updates, and resource sharing. It allows developers to spin up specialized agents—such as marketing assistants or research bots—without manually editing sensitive JSON files or setting up boilerplate environments from scratch.
This skill ensures that every new addition to your team is perfectly integrated into the communication gateway while remaining securely isolated in its own workspace. By utilizing Openclaw Skills for this process, you ensure consistent application of security policies, authentication profiles, and tool permissions across your entire multi-agent architecture.
Add Agent Use Cases
- Deploying specialized AI assistants with unique Telegram bot identities for specific business functions.
- Rapidly scaling multi-agent systems by cloning skills and user preferences from a primary instance.
- Automating the setup of isolated development or staging environments for new agent logic.
- Configuring complex inter-agent communication channels to allow bots to collaborate on shared tasks.
How Add Agent Works
- Extract configuration parameters like Agent ID, Name, Telegram Bot Token, and persona descriptions from natural language input.
- Detect the existing installation environment to determine the correct paths for new workspaces and agent directories.
- Create a secure, timestamped backup of the openclaw.json file to prevent data loss during the update process.
- Initialize the agent directory structure and default persona files using core CLI commands.
- Synchronize essential assets including authentication profiles, existing skills, and user context files from the main agent.
- Generate tailored SOUL.md and AGENTS.md files that define the new agent's identity and collaboration protocols.
- Update the global configuration to include new agent bindings, telegram channel accounts, and inter-agent communication permissions.
- Perform a JSON validation check and correct file system permissions to ensure the environment is ready for execution.
Add Agent Setup
To deploy this skill within your Openclaw Skills environment, ensure the Openclaw CLI is active and you have a primary agent configured.
Execution
Trigger the process by providing the agent details in a natural language prompt:
Add a new agent, ID is research-bot, name is Researcher, telegram Bot Token is YOUR_TOKEN, allowFrom is YOUR_ID, responsible for web research
Finalizing Setup
Once the automated steps complete, you must manually restart the gateway to apply the new configuration:
openclaw gateway restart
Verify the new agent is correctly bound:
openclaw agents list --bindings
Add Agent Data Schema & Taxonomy
The Add Agent skill manages a structured hierarchy within your state directory to ensure modularity and ease of maintenance:
| Component | Location | Purpose |
|---|---|---|
| Workspace | ${STATE_DIR}/workspace-${AGENT_ID} |
Contains the agent's unique skills, memory, and persona files. |
| Agent Dir | ${STATE_DIR}/agents/${AGENT_ID}/agent |
Stores logic, authentication profiles, and core configuration. |
| SOUL.md | Workspace Root | Defines the identity, personality, and core rules of the specific agent. |
| AGENTS.md | Workspace Root | Details the collaboration rules for interacting with the main agent. |
| openclaw.json | Config Path | The central manifest updated with new bindings and tool permissions. |
Add Agent Advanced Features
- Automated rollback mechanism that restores the previous configuration if JSON validation fails.
- Dynamic permission management that automatically appends new agents to the agentToAgent communication allowlist.
- Permission inheritance that ensures new agent workspaces match the ownership and group settings of the main environment.
- Contextual persona generation which writes custom identity and responsibility blocks based on user-provided descriptions.
- Seamless skill synchronization that clones your library of Openclaw Skills to every new agent instance automatically.
SKILL.md
Loading
Related Openclaw Skills
Bitkit CLI
A powerful Bitcoin Lightning payment CLI and self-custodial wallet designed specifically for AI agents and automated workflows.
Immortal Brain: Proactive Autonomous AI Agent
A highly autonomous AI agent that independently manages the full lifecycle of tasks through proactive research, planning, and execution.
Tesla Commands
A specialized tool for controlling Tesla vehicle climate, charging limits, and schedules through the MyTeslaMate API.
Nordpool Finland Energy Prices
A specialized tool for fetching real-time Finnish electricity prices and identifying the cheapest windows for EV charging.
Seedream Image Generation
A high-performance AI image generation tool for Openclaw Skills using the Seedream API and Doubao-Seedream models.
Global Temperature Bucket Thresholds (Simmer)
An automated Polymarket trading strategy that manages Global Temperature bucket thresholds based on specific probability triggers.