aitbc/.windsurf/skills/openclaw-agent-communicator.md

description, title, version

description	title	version
Atomic OpenClaw agent communication with deterministic message handling and response validation	openclaw-agent-communicator	1.0

OpenClaw Agent Communicator

Purpose

Handle OpenClaw agent message delivery, response processing, and communication validation with deterministic outcome tracking.

Activation

Trigger when user requests agent communication: message sending, response analysis, or communication validation.

Input

{
  "operation": "send|receive|analyze|validate",
  "agent": "main|specific_agent_name",
  "message": "string (for send)",
  "session_id": "string (optional for send/validate)",
  "thinking_level": "off|minimal|low|medium|high|xhigh",
  "response": "string (for receive/analyze)",
  "expected_response": "string (optional for validate)",
  "timeout": "number (optional, default 30 seconds)",
  "context": "string (optional for send)"
}

Output

{
  "summary": "Agent communication operation completed successfully",
  "operation": "send|receive|analyze|validate",
  "agent": "string",
  "session_id": "string",
  "message": "string (for send)",
  "response": "string (for receive/analyze)",
  "thinking_level": "string",
  "response_time": "number",
  "response_quality": "number (0-1)",
  "context_preserved": "boolean",
  "communication_issues": [],
  "recommendations": [],
  "confidence": 1.0,
  "execution_time": "number",
  "validation_status": "success|partial|failed"
}

Process

1. Analyze

• Validate agent availability
• Check message format and content
• Verify thinking level compatibility
• Assess communication requirements

2. Plan

• Prepare message parameters
• Set session management strategy
• Define response validation criteria
• Configure timeout handling

3. Execute

• Execute OpenClaw agent command
• Capture agent response
• Measure response time
• Analyze response quality

4. Validate

• Verify message delivery success
• Check response completeness
• Validate context preservation
• Assess communication effectiveness

Constraints

• MUST NOT send messages to unavailable agents
• MUST NOT exceed message length limits (4000 characters)
• MUST validate thinking level compatibility
• MUST handle communication timeouts gracefully
• MUST preserve session context when specified
• MUST validate response format and content

Environment Assumptions

• OpenClaw 2026.3.24+ installed and gateway running
• Agent workspace configured at ~/.openclaw/workspace/
• Network connectivity for agent communication
• Default agent available: "main"
• Session management functional

Error Handling

• Agent unavailable → Return agent status and availability recommendations
• Communication timeout → Return timeout details and retry suggestions
• Invalid thinking level → Return valid thinking level options
• Message too long → Return truncation recommendations

Example Usage Prompt

Send message to main agent with medium thinking level: "Analyze the current blockchain status and provide optimization recommendations for better performance"

Expected Output Example

{
  "summary": "Message sent to main agent successfully with comprehensive blockchain analysis response",
  "operation": "send",
  "agent": "main",
  "session_id": "session_1774883100",
  "message": "Analyze the current blockchain status and provide optimization recommendations for better performance",
  "response": "Current blockchain status: Chain height 12345, active nodes 2, block time 15s. Optimization recommendations: 1) Increase block size for higher throughput, 2) Implement transaction batching, 3) Optimize consensus algorithm for faster finality.",
  "thinking_level": "medium",
  "response_time": 8.5,
  "response_quality": 0.9,
  "context_preserved": true,
  "communication_issues": [],
  "recommendations": ["Consider implementing suggested optimizations", "Monitor blockchain performance after changes", "Test optimizations in staging environment"],
  "confidence": 1.0,
  "execution_time": 8.7,
  "validation_status": "success"
}

Model Routing Suggestion

Fast Model (Claude Haiku, GPT-3.5-turbo)

• Simple message sending with low thinking
• Basic response validation
• Communication status checking

Reasoning Model (Claude Sonnet, GPT-4)

• Complex message sending with high thinking
• Response analysis and quality assessment
• Communication optimization recommendations
• Error diagnosis and recovery

Performance Notes

• Execution Time: 1-3 seconds for simple messages, 5-15 seconds for complex analysis
• Memory Usage: <100MB for agent communication
• Network Requirements: OpenClaw gateway connectivity
• Concurrency: Safe for multiple simultaneous agent communications
• Session Management: Automatic context preservation across multiple messages