cloud-workflow（云工作流程）

Cloud Workflow

Design and orchestrate event-driven automation workflows with intelligent agent coordination and message queue processing.

Quick Start

// Create a CI/CD workflow
const workflow = await mcp__flow-nexus__workflow_create({
  name: "CI/CD Pipeline",
  description: "Automated testing and deployment",
  steps: [
    { id: "test", action: "run_tests", agent: "tester" },
    { id: "build", action: "build_app", agent: "builder" },
    { id: "deploy", action: "deploy_prod", agent: "deployer" }
  ],
  triggers: ["push_to_main", "manual_trigger"]
});

// Execute the workflow
await mcp__flow-nexus__workflow_execute({
  workflow_id: workflow.workflow_id,
  input_data: { branch: "main" },
  async: true
});

When to Use

• Automating CI/CD pipelines with multiple stages
• Orchestrating data processing and ETL workflows
• Creating event-driven automation for business processes
• Managing multi-stage review and approval workflows
• Scheduling recurring automated tasks
• Coordinating complex multi-agent collaboration

Prerequisites

• Flow Nexus account with active session
• MCP server flow-nexus configured
• Sufficient rUv credits for workflow execution

Core Concepts

Workflow Patterns

Execution Strategies

• Sequential: Steps run one after another
• Parallel: Independent steps run simultaneously
• Conditional: Steps execute based on conditions

Agent Assignment

Workflows can automatically assign optimal agents to tasks using:

• Explicit Assignment: Specify agent type per step
• Vector Similarity: AI-powered matching based on task requirements

MCP Tools Reference

Workflow Creation

mcp__flow-nexus__workflow_create({
  name: "Workflow Name",
  description: "Workflow description",
  steps: [
    {
      id: "step1",
      action: "action_name",
      agent: "agent_type",     // Optional: auto-assigned if not specified
      config: {}               // Step-specific configuration
    },
    {
      id: "step2",
      action: "action_name",
      depends: ["step1"]       // Dependencies on other steps
    }
  ],
  triggers: ["trigger1", "trigger2"],  // Event triggers
  priority: 5,                 // Priority 0-10
  metadata: {}                 // Additional metadata
})
// Returns: { workflow_id, name, status, created_at }

Workflow Execution

mcp__flow-nexus__workflow_execute({
  workflow_id: "workflow_id",
  input_data: {                // Input data for execution
    key: "value"
  },
  async: true                  // Execute asynchronously via queue
})
// Returns: { execution_id, status, started_at }

Status and Monitoring

// Get workflow status
mcp__flow-nexus__workflow_status({
  workflow_id: "workflow_id",
  execution_id: "execution_id",  // Optional: specific execution
  include_metrics: true
})
// Returns: { status, progress, metrics, step_results }

// List all workflows
mcp__flow-nexus__workflow_list({
  status: "active",            // Filter by status
  limit: 10,
  offset: 0
})

// Check message queue status
mcp__flow-nexus__workflow_queue_status({
  queue_name: "queue_name",    // Optional: specific queue
  include_messages: true
})

Agent Assignment

mcp__flow-nexus__workflow_agent_assign({
  task_id: "task_id",
  agent_type: "coder",         // Preferred agent type
  use_vector_similarity: true  // Use AI matching
})
// Returns: { agent_id, type, match_score }

Audit Trail

mcp__flow-nexus__workflow_audit_trail({
  workflow_id: "workflow_id",
  start_time: "2026-01-01T00:00:00Z",
  limit: 50
})
// Returns: { events: [{ timestamp, action, user, details }] }

Usage Examples

Example 1: CI/CD Pipeline

// Create comprehensive CI/CD workflow
const cicdWorkflow = await mcp__flow-nexus__workflow_create({
  name: "Full CI/CD Pipeline",
  description: "Complete testing, building, and deployment workflow",
  steps: [
    {
      id: "lint",
      action: "run_linter",
      agent: "code-analyzer",
      config: { strict: true }
    },
    {
      id: "test",
      action: "run_tests",
      agent: "tester",
      config: { coverage_threshold: 80 },
      depends: ["lint"]
    },
    {
      id: "security_scan",
      action: "security_check",
      agent: "security-analyzer",
      depends: ["lint"]
    },
    {
      id: "build",
      action: "build_app",
      agent: "builder",
      depends: ["test", "security_scan"]
    },
    {
      id: "deploy_staging",
      action: "deploy",
      agent: "deployer",
      config: { environment: "staging" },
      depends: ["build"]
    },
    {
      id: "integration_tests",
      action: "run_integration_tests",
      agent: "tester",
      depends: ["deploy_staging"]
    },
    {
      id: "deploy_prod",
      action: "deploy",
      agent: "deployer",
      config: { environment: "production" },
      depends: ["integration_tests"]
    }
  ],
  triggers: ["push_to_main", "release_tag"],
  priority: 8
});

// Execute on push
await mcp__flow-nexus__workflow_execute({
  workflow_id: cicdWorkflow.workflow_id,
  input_data: {
    branch: "main",
    commit: "abc123",
    author: "developer@example.com"
  },
  async: true
});

// Monitor progress
const status = await mcp__flow-nexus__workflow_status({
  workflow_id: cicdWorkflow.workflow_id,
  include_metrics: true
});

console.log(`Progress: ${status.progress}%, Current step: ${status.current_step}`);

Example 2: Data Processing Pipeline

// ETL workflow with validation
const etlWorkflow = await mcp__flow-nexus__workflow_create({
  name: "Data ETL Pipeline",
  description: "Extract, transform, and load data with validation",
  steps: [
    {
      id: "extract",
      action: "extract_data",
      agent: "data-extractor",
      config: { source: "s3://bucket/raw-data" }
    },
    {
      id: "validate",
      action: "validate_schema",
      agent: "data-validator",
      depends: ["extract"]
    },
    {
      id: "transform",
      action: "transform_data",
      agent: "data-transformer",
      config: { rules: ["normalize", "dedupe", "enrich"] },
      depends: ["validate"]
    },
    {
      id: "quality_check",
      action: "run_quality_checks",
      agent: "data-analyst",
      depends: ["transform"]
    },
    {
      id: "load",
      action: "load_to_warehouse",
      agent: "data-loader",
      config: { target: "postgres://warehouse" },
      depends: ["quality_check"]
    }
  ],
  triggers: ["schedule:0 2 * * *", "manual_trigger"]  // Daily at 2 AM
});

// Manual execution
await mcp__flow-nexus__workflow_execute({
  workflow_id: etlWorkflow.workflow_id,
  input_data: { date: "2026-01-02" }
});

Example 3: Multi-Stage Code Review

// Automated code review workflow
const reviewWorkflow = await mcp__flow-nexus__workflow_create({
  name: "Automated Code Review",
  description: "Multi-stage code analysis and review",
  steps: [
    {
      id: "static_analysis",
      action: "run_static_analysis",
      agent: "code-analyzer"
    },
    {
      id: "security_review",
      action: "security_scan",
      agent: "security-reviewer",
      depends: ["static_analysis"]
    },
    {
      id: "performance_review",
      action: "analyze_performance",
      agent: "perf-analyzer",
      depends: ["static_analysis"]
    },
    {
      id: "ai_review",
      action: "ai_code_review",
      agent: "ai-reviewer",
      depends: ["static_analysis"]
    },
    {
      id: "compile_report",
      action: "generate_report",
      agent: "report-generator",
      depends: ["security_review", "performance_review", "ai_review"]
    }
  ],
  triggers: ["pull_request_opened", "pull_request_updated"]
});

// Assign optimal agent dynamically
await mcp__flow-nexus__workflow_agent_assign({
  task_id: "security_review_123",
  use_vector_similarity: true
});

Example 4: Queue Management

// Check queue status
const queueStatus = await mcp__flow-nexus__workflow_queue_status({
  include_messages: true
});

console.log(`Pending messages: ${queueStatus.pending}`);
console.log(`Processing: ${queueStatus.processing}`);

// Review audit trail
const audit = await mcp__flow-nexus__workflow_audit_trail({
  workflow_id: "workflow_id",
  limit: 100
});

for (const event of audit.events) {
  console.log(`${event.timestamp}: ${event.action} by ${event.user}`);
}

Execution Checklist

• Define workflow steps and dependencies
• Assign or auto-assign agents to steps
• Configure triggers (events, schedules)
• Set workflow priority
• Create the workflow
• Execute with appropriate input data
• Monitor progress and step status
• Review audit trail for compliance
• Clean up or archive completed workflows

Best Practices

1. Step Granularity: Break complex tasks into atomic steps for better monitoring
2. Dependency Chains: Carefully plan dependencies to maximize parallelism
3. Error Handling: Include retry logic and fallback steps
4. Async Execution: Use async mode for long-running workflows
5. Agent Matching: Leverage vector similarity for optimal agent assignment
6. Audit Compliance: Regularly review audit trails for security

Error Handling

Metrics & Success Criteria

• Workflow Completion Rate: Target >95%
• Average Execution Time: Track per workflow type
• Queue Latency: <10 seconds for async jobs
• Agent Utilization: >80% during active workflows
• Error Rate: <5% per workflow type

Integration Points

With Swarms

// Swarm-powered workflow
const swarm = await mcp__flow-nexus__swarm_init({ topology: "mesh" });

await mcp__flow-nexus__workflow_create({
  name: "Swarm Workflow",
  steps: [
    { id: "task", action: "swarm_execute", config: { swarm_id: swarm.swarm_id } }
  ]
});

With Sandboxes

// Sandbox execution in workflow
await mcp__flow-nexus__workflow_create({
  name: "Sandbox Pipeline",
  steps: [
    { id: "create", action: "sandbox_create", config: { template: "node" } },
    { id: "test", action: "sandbox_execute", depends: ["create"] },
    { id: "cleanup", action: "sandbox_delete", depends: ["test"] }
  ]
});

Related Skills

• cloud-swarm - Multi-agent orchestration
• cloud-sandbox - Isolated execution environments
• cloud-neural - Neural network operations

References

• Flow Nexus Workflow Documentation
• Event-Driven Architecture Patterns

Version History

• 1.0.0 (2026-01-02): Initial release - converted from flow-nexus-workflow agent