# Orchestrate Multi-Agent AI Swarms

Complex AI work can stall when one agent handles every task. This skill guides Claude, Codex, and Claude Code through agentic-flow swarm patterns for coordinated parallel work.

## Install

```bash
npx skillstore add dnyoussef/swarm-orchestration
```

## Metadata

- Status: approved
- Slug: dnyoussef-swarm-orchestration
- Skillstore revision: r1
- Version status: missing
- Tree hash: 2fafc00e741a088759dfbd455ecb32aa477e4e08b43db0a9e651631072a43bb2
- Author: DNYoussef
- GitHub username: DNYoussef
- License: MIT
- Repository: https://github.com/DNYoussef/ai-chrome-extension/tree/main/.claude/skills/swarm-orchestration
- Ref: 0519034dad657fb1f7706e0550e962beeda73fdf
- 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
- Quality score: 78
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/dnyoussef-swarm-orchestration
- Manifest: https://skillstore.pages.dev/api/skills/dnyoussef-swarm-orchestration/manifest

## Capabilities

- Explains mesh, hierarchical, and adaptive swarm topologies for coordinated agent work.
- Shows how to initialize swarms, spawn agents, and orchestrate parallel tasks with agentic-flow.
- Defines pipeline patterns with staged dependencies for design, implementation, testing, and review.
- Describes shared memory coordination so agents can store and retrieve project state.
- Covers load balancing, retry behavior, fallback assignment, performance metrics, and hook integration.

## Use Cases

- Plan Parallel Feature Delivery: Split a software feature across coding, testing, review, and architecture agents with clear coordination points.
- Coordinate Complex Automation: Choose a swarm topology and hook flow for workflows that need shared state and staged execution.
- Improve Multi-Agent Reliability: Add retry, fallback, load balancing, and monitoring patterns to reduce stalled or uneven agent work.

## Prompt Templates

### Start a Small Swarm

```
Use swarm-orchestration to design a three-agent setup for [project]. Include topology, agent roles, and the first task.
```

### Plan Parallel Delivery

```
Use swarm-orchestration to split [feature] across coding, testing, and review agents. Include dependencies, timeouts, and coordination points.
```

### Add Shared Memory

```
Use swarm-orchestration to design shared memory keys for [workflow]. Include what each agent writes, reads, and validates.
```

### Optimize a Resilient Swarm

```
Use swarm-orchestration to compare mesh, hierarchical, and adaptive topologies for [system]. Recommend load balancing, failure handling, and monitoring.
```

## Limitations

- Requires agentic-flow v1.5.11 or newer and Node.js 18 or newer.
- Provides orchestration guidance and examples, not the agentic-flow runtime itself.
- Does not validate task safety, package trust, or workspace permissions automatically.
- External npx examples should be reviewed and pinned before use in sensitive environments.

## Best Practices

- Start with two or three agents before increasing swarm size.
- Use shared memory for schemas, decisions, status, and unresolved risks.
- Set timeouts, monitor metrics, and define fallback behavior before long runs.

## Anti Patterns

- Starting many agents without clear roles, ownership, or task boundaries.
- Using parallel execution when tasks have unresolved dependencies.
- Running external hook commands without reviewing package source and workspace impact.

## Security Audit

- Audited at: 2026-07-09T12:04:54.115\+00:00
- Summary: I found no prompt injection, data exfiltration, or malicious social-engineering language in SKILL.md. Most static Ruby/shell findings are Markdown fence false positives; two npx bash examples are confirmed medium-risk external command guidance.

## Stats

- Views: 255
- Downloads: 13
- Favorites: 0
- Popularity score: 0