clawhub-skill-remote-agent

Bridge to external vertical agents (Google ADK, VeADK, etc.) for specialized tasks.

Remote Agent Bridge

This skill enables OpenClaw to delegate tasks to external, specialized AI agents via a standard HTTP interface. Use this when the user's request requires domain-specific knowledge (e.g., enterprise data, financial analysis, legal review) that is handled by a separate agent system.

Configuration

Ensure the following environment variables are set in your OpenClaw environment (e.g., via .env or openclaw config):

• REMOTE_AGENT_URL: The HTTP endpoint of the external agent (e.g., https://remote-agent.example.com/run or your Google ADK endpoint).
• REMOTE_AGENT_KEY: (Optional) The Bearer token for authentication.

Usage

When the user asks a question that falls into the domain of a specialized remote agent, use this skill to forward the request.

Command

bash

python3 skills/remote-agent/scripts/client.py --query "<USER_QUERY>" [--agent "<AGENT_ID>"]

Examples

Scenario 1: Financial Analysis (VeADK)

User: "Analyze the Q3 earnings report for TechCorp." Thought: The user is asking for financial analysis. I should delegate this to the 'financial-expert' agent. Action:

bash

python3 skills/remote-agent/scripts/client.py --agent "financial-expert" --query "Analyze the Q3 earnings report for TechCorp"

Scenario 2: Enterprise Knowledge (Google ADK)

User: "What is the company policy on remote work?" Thought: This requires internal knowledge. I'll ask the 'hr-bot'. Action:

bash

python3 skills/remote-agent/scripts/client.py --agent "hr-bot" --query "company policy on remote work"

Scenario 3: Custom LangChain Backend

User: "Run the data processing pipeline." Action:

bash

python3 skills/remote-agent/scripts/client.py --query "Run the data processing pipeline"