# Orchestrate Cloud Agent Swarms and Workflows

Complex AI projects require coordinated agents, dependencies, queues, and monitoring. This skill guides Flow Nexus swarm creation, workflow execution, scaling, templates, and audit tracking.

## Install

```bash
npx skillstore add ruvnet/flow-nexus-swarm
```

## Metadata

- Status: approved
- Slug: ruvnet-flow-nexus-swarm
- Skillstore revision: r2
- Version status: missing
- Tree hash: a7c1a4fae507bf0bb8d2d20f6ac4715868307895118abf2efbb5128479501389
- Author: ruvnet
- GitHub username: ruvnet
- License: MIT
- Repository: https://github.com/ruvnet/claude-flow/tree/main/.claude/skills/flow-nexus-swarm
- Ref: ebdfe608f5de2b66ff37ab4af12af8ac4f5e8006
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, network
- Quality score: 69
- Public page: https://skillstore.pages.dev/skills/ruvnet-flow-nexus-swarm
- Manifest: https://skillstore.pages.dev/api/skills/ruvnet-flow-nexus-swarm/manifest

## Capabilities

- Configures hierarchical, mesh, ring, and star swarm topologies through Flow Nexus MCP tools.
- Spawns specialized researcher, coder, analyst, optimizer, and coordinator agents.
- Creates workflows with dependencies, parallel steps, triggers, priorities, retries, and failure handling.
- Executes workflows synchronously or through asynchronous message queues.
- Monitors swarm status, workflow metrics, audit trails, queues, execution streams, and generated files.
- Scales or destroys swarms and creates configurations from reusable templates.

## Use Cases

- Coordinate Multi-Repository Delivery: Create specialized agent roles and dependency-aware workflows for synchronized development, testing, and deployment across repositories.
- Run Parallel Research: Deploy a mesh swarm of researchers and analysts to gather evidence, compare sources, and synthesize findings.
- Automate Data Pipelines: Model extraction, validation, transformation, enrichment, loading, retries, and scheduled execution as one monitored workflow.

## Prompt Templates

### Create a Basic Swarm

```
Create a Flow Nexus swarm for [task]. Recommend a topology, strategy, agent count, and specialized roles. Explain each choice before deployment.
```

### Design a Workflow

```
Design a Flow Nexus workflow for [process]. Define steps, dependencies, parallel work, triggers, retries, timeouts, and success criteria.
```

### Operate and Scale a Swarm

```
Review the active swarm for [project]. Summarize status and workload, recommend scaling changes, identify failed work, and preserve an audit trail.
```

### Orchestrate a Multi-Phase Program

```
Plan coordinated Flow Nexus swarms for research, implementation, testing, and deployment. Define handoffs, shared inputs, failure recovery, approvals, metrics, and cleanup.
```

## Limitations

- Requires a Flow Nexus account, network access, and compatible CLI or MCP configuration.
- Does not implement the Flow Nexus service or guarantee that documented tools remain available.
- Cloud execution may expose task inputs, code, or generated files to external infrastructure.
- Examples require adaptation for project permissions, budgets, error policies, and production controls.

## Best Practices

- Choose topology from task dependencies and communication needs before creating agents.
- Define retries, timeouts, failure actions, approvals, and cleanup for every production workflow.
- Review status, metrics, queues, and audit trails before scaling or changing active work.

## Anti Patterns

- Do not assign every task to a large swarm without clear roles or dependency boundaries.
- Do not place secrets or regulated data in cloud workflow inputs without approved controls.
- Do not execute unpinned latest package versions in trusted or production environments.

## Security Audit

- Audited at: 2026-07-23T18:50:16.662\+00:00
- Summary: All 66 backtick findings are Markdown fences or inline code notation, not Ruby or shell execution. The three URLs are support resources, and the hook is not reconnaissance. Setup still introduces medium supply-chain risk by executing unpinned packages labeled latest.

## Stats

- Views: 189
- Downloads: 9
- Favorites: 0
- Popularity score: 0