aitbc/.windsurf/workflows/ollama-gpu-test-openclaw.md

description, title, version

description	title	version
OpenClaw agent workflow for complete Ollama GPU provider testing from client submission to blockchain recording	OpenClaw Ollama GPU Provider Test Workflow	1.0

OpenClaw Ollama GPU Provider Test Workflow

This OpenClaw agent workflow executes the complete end-to-end test for Ollama GPU inference jobs, including payment processing and blockchain transaction recording.

Prerequisites

• OpenClaw 2026.3.24+ installed and gateway running
• All services running: coordinator, GPU miner, Ollama, blockchain node
• Home directory wallets configured
• Enhanced CLI with multi-wallet support

Agent Roles

Test Coordinator Agent

Purpose: Orchestrate the complete Ollama GPU test workflow

• Coordinate test execution across all services
• Monitor progress and validate results
• Handle error conditions and retry logic

Client Agent

Purpose: Simulate client submitting AI inference jobs

• Create and manage test wallets
• Submit inference requests to coordinator
• Monitor job progress and results

Miner Agent

Purpose: Simulate GPU provider processing jobs

• Monitor GPU miner service status
• Track job processing and resource utilization
• Validate receipt generation and pricing

Blockchain Agent

Purpose: Verify blockchain transaction recording

• Monitor blockchain for payment transactions
• Validate transaction confirmations
• Check wallet balance updates

OpenClaw Agent Workflow

Phase 1: Environment Validation

# Initialize test coordinator
SESSION_ID="ollama-test-$(date +%s)"
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Initialize Ollama GPU provider test workflow. Validate all services and dependencies." \
    --thinking high

# Agent performs environment checks
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Execute environment validation: check coordinator API, Ollama service, GPU miner, blockchain node health" \
    --thinking medium

Phase 2: Wallet Setup

# Initialize client agent
openclaw agent --agent client-agent --session-id $SESSION_ID \
    --message "Initialize as client agent. Create test wallets and configure for AI job submission." \
    --thinking medium

# Agent creates test wallets
openclaw agent --agent client-agent --session-id $SESSION_ID \
    --message "Create test wallets: test-client and test-miner. Switch to client wallet and verify balance." \
    --thinking medium \
    --parameters "wallet_type:simple,backup_enabled:true"

# Initialize miner agent
openclaw agent --agent miner-agent --session-id $SESSION_ID \
    --message "Initialize as miner agent. Verify miner wallet and GPU resource availability." \
    --thinking medium

Phase 3: Service Health Verification

# Coordinator agent checks all services
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Perform comprehensive service health check: coordinator API, Ollama GPU service, GPU miner service, blockchain RPC" \
    --thinking high \
    --parameters "timeout:30,retry_count:3"

# Agent reports service status
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Report service health status and readiness for GPU testing" \
    --thinking medium

Phase 4: GPU Test Execution

# Client agent submits inference job
openclaw agent --agent client-agent --session-id $SESSION_ID \
    --message "Submit Ollama GPU inference job: 'What is the capital of France?' using llama3.2:latest model" \
    --thinking high \
    --parameters "prompt:What is the capital of France?,model:llama3.2:latest,payment:10"

# Agent monitors job progress
openclaw agent --agent client-agent --session-id $SESSION_ID \
    --message "Monitor job progress through states: QUEUED → RUNNING → COMPLETED" \
    --thinking medium \
    --parameters "polling_interval:5,timeout:300"

# Agent validates job results
openclaw agent --agent client-agent --session-id $SESSION_ID \
    --message "Validate job result: 'The capital of France is Paris.' Check accuracy and completeness" \
    --thinking medium

Phase 5: Payment Processing

# Client agent handles payment processing
openclaw agent --agent client-agent --session-id $SESSION_ID \
    --message "Process payment for completed GPU job: verify receipt information, pricing, and total cost" \
    --thinking high \
    --parameters "validate_receipt:true,check_pricing:true"

# Agent reports payment details
openclaw agent --agent client-agent --session-id $SESSION_ID \
    --message "Report payment details: receipt ID, provider, GPU seconds, unit price, total cost" \
    --thinking medium

Phase 6: Blockchain Verification

# Blockchain agent verifies transaction recording
openclaw agent --agent blockchain-agent --session-id $SESSION_ID \
    --message "Verify blockchain transaction recording: check for payment transaction, validate confirmation, track block inclusion" \
    --thinking high \
    --parameters "confirmations:1,timeout:60"

# Agent reports blockchain status
openclaw agent --agent blockchain-agent --session-id $SESSION_ID \
    --message "Report blockchain verification results: transaction hash, block height, confirmation status" \
    --thinking medium

Phase 7: Final Balance Verification

# Client agent checks final wallet balances
openclaw agent --agent client-agent --session-id $SESSION_ID \
    --message "Verify final wallet balances after transaction: compare initial vs final balances" \
    --thinking medium

# Miner agent checks earnings
openclaw agent --agent miner-agent --session-id $SESSION_ID \
    --message "Verify miner earnings: check wallet balance increase from GPU job payment" \
    --thinking medium

Phase 8: Test Completion

# Coordinator agent generates final report
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Generate comprehensive test completion report: all phases status, results, wallet changes, blockchain verification" \
    --thinking xhigh \
    --parameters "include_metrics:true,include_logs:true,format:comprehensive"

# Agent posts results to coordination topic
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Post test results to blockchain coordination topic for permanent recording" \
    --thinking high

OpenClaw Agent Templates

Test Coordinator Agent Template

{
  "name": "Ollama Test Coordinator",
  "type": "test-coordinator",
  "description": "Coordinates complete Ollama GPU provider test workflow",
  "capabilities": ["orchestration", "monitoring", "validation", "reporting"],
  "configuration": {
    "timeout": 300,
    "retry_count": 3,
    "validation_strict": true
  }
}

Client Agent Template

{
  "name": "AI Test Client",
  "type": "client-agent", 
  "description": "Simulates client submitting AI inference jobs",
  "capabilities": ["wallet_management", "job_submission", "payment_processing"],
  "configuration": {
    "default_model": "llama3.2:latest",
    "default_payment": 10,
    "wallet_type": "simple"
  }
}

Miner Agent Template

{
  "name": "GPU Test Miner",
  "type": "miner-agent",
  "description": "Monitors GPU provider and validates job processing",
  "capabilities": ["resource_monitoring", "receipt_validation", "earnings_tracking"],
  "configuration": {
    "monitoring_interval": 10,
    "gpu_utilization_threshold": 0.8
  }
}

Blockchain Agent Template

{
  "name": "Blockchain Verifier",
  "type": "blockchain-agent",
  "description": "Verifies blockchain transactions and confirmations",
  "capabilities": ["transaction_monitoring", "balance_tracking", "confirmation_verification"],
  "configuration": {
    "confirmations_required": 1,
    "monitoring_interval": 15
  }
}

Expected Test Results

Success Indicators

✅ Environment Check: All services healthy
✅ Wallet Setup: Test wallets created and funded
✅ Service Health: Coordinator, Ollama, GPU miner, blockchain operational
✅ GPU Test: Job submitted and completed successfully
✅ Payment Processing: Receipt generated and validated
✅ Blockchain Recording: Transaction found and confirmed
✅ Balance Verification: Wallet balances updated correctly

Key Metrics

💰 Initial Wallet Balances:
   Client: 9365.0 AITBC
   Miner:  1525.0 AITBC

📤 Job Submission:
   Prompt: What is the capital of France?
   Model: llama3.2:latest
   Payment: 10 AITBC

📊 Job Result:
   Output: The capital of France is Paris.

🧾 Payment Details:
   Receipt ID: receipt_123
   Provider: miner_dev_key_1
   GPU Seconds: 45
   Unit Price: 0.02 AITBC
   Total Price: 0.9 AITBC

⛓️ Blockchain Verification:
   TX Hash: 0xabc123...
   Block: 12345
   Confirmations: 1

💰 Final Wallet Balances:
   Client: 9364.1 AITBC (-0.9 AITBC)
   Miner:  1525.9 AITBC (+0.9 AITBC)

Error Handling

Common Issues and Agent Responses

# Service Health Issues
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Service health check failed. Implementing recovery procedures: restart services, verify connectivity, check logs" \
    --thinking high

# Wallet Issues
openclaw agent --agent client-agent --session-id $SESSION_ID \
    --message "Wallet operation failed. Implementing wallet recovery: check keystore, verify permissions, recreate wallet if needed" \
    --thinking high

# GPU Issues
openclaw agent --agent miner-agent --session-id $SESSION_ID \
    --message "GPU processing failed. Implementing recovery: check GPU availability, restart Ollama, verify model availability" \
    --thinking high

# Blockchain Issues
openclaw agent --agent blockchain-agent --session-id $SESSION_ID \
    --message "Blockchain verification failed. Implementing recovery: check node sync, verify transaction pool, retry with different parameters" \
    --thinking high

Performance Monitoring

Agent Performance Metrics

# Monitor agent performance
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Report agent performance metrics: response time, success rate, error count, resource utilization" \
    --thinking medium

# System performance during test
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Monitor system performance during GPU test: CPU usage, memory usage, GPU utilization, network I/O" \
    --thinking medium

OpenClaw Integration

Session Management

# Create persistent session for entire test
SESSION_ID="ollama-gpu-test-$(date +%s)"

# Use session across all agents
openclaw agent --agent test-coordinator --session-id $SESSION_ID --message "Initialize test" --thinking high
openclaw agent --agent client-agent --session-id $SESSION_ID --message "Submit job" --thinking medium
openclaw agent --agent miner-agent --session-id $SESSION_ID --message "Monitor GPU" --thinking medium
openclaw agent --agent blockchain-agent --session-id $SESSION_ID --message "Verify blockchain" --thinking high

Cross-Agent Communication

# Agents communicate through coordination topic
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Post coordination message: Test phase completed, next phase starting" \
    --thinking medium

# Other agents respond to coordination
openclaw agent --agent client-agent --session-id $SESSION_ID \
    --message "Acknowledge coordination: Ready for next phase" \
    --thinking minimal

Automation Script

Complete Test Automation

#!/bin/bash
# ollama_gpu_test_openclaw.sh

SESSION_ID="ollama-gpu-test-$(date +%s)"

echo "Starting OpenClaw Ollama GPU Provider Test..."

# Initialize coordinator
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Initialize complete Ollama GPU test workflow" \
    --thinking high

# Execute all phases automatically
openclaw agent --agent test-coordinator --session-id $SESSION_ID \
    --message "Execute complete test: environment check, wallet setup, service health, GPU test, payment processing, blockchain verification, final reporting" \
    --thinking xhigh \
    --parameters "auto_execute:true,timeout:600,report_format:comprehensive"

echo "OpenClaw Ollama GPU test completed!"

Integration with Existing Workflow

From Manual to Automated

# Manual workflow (original)
cd /home/oib/windsurf/aitbc/home
python3 test_ollama_blockchain.py

# OpenClaw automated workflow
./ollama_gpu_test_openclaw.sh

Benefits of OpenClaw Integration

• Intelligent Error Handling: Agents detect and recover from failures
• Adaptive Testing: Agents adjust test parameters based on system state
• Comprehensive Reporting: Agents generate detailed test reports
• Cross-Node Coordination: Agents coordinate across multiple nodes
• Blockchain Recording: Results permanently recorded on blockchain

Troubleshooting

Agent Communication Issues

# Check OpenClaw gateway status
openclaw status --agent all

# Test agent communication
openclaw agent --agent test --message "ping" --thinking minimal

# Check session context
openclaw agent --agent test-coordinator --session-id $SESSION_ID --message "report status" --thinking medium

Service Integration Issues

# Verify service endpoints
curl -s http://localhost:11434/api/tags
curl -s http://localhost:8006/health
systemctl is-active aitbc-host-gpu-miner.service

# Test CLI integration
./aitbc-cli --help
./aitbc-cli wallet info

This OpenClaw agent workflow transforms the manual Ollama GPU test into an intelligent, automated, and blockchain-recorded testing process with comprehensive error handling and reporting capabilities.