# Analyze Multi-Agent Coordination

Multi-agent systems can hide brittle handoffs, shared-state coupling, and unclear communication paths. This skill guides Claude, Codex, or Claude Code to map coordination models and produce a structured review.

## Install

```bash
npx skillstore add dowwie/multi-agent-analysis
```

## Metadata

- Status: approved
- Slug: dowwie-multi-agent-analysis
- Skillstore revision: r1
- Version status: missing
- Tree hash: d7bfa8594e102835fbfb05001fcb00a0ecc1cd0de6f09329ea69f59876bfef07
- Author: Dowwie
- GitHub username: Dowwie
- License: MIT
- Repository: https://github.com/Dowwie/agent\_framework\_study/tree/main/.claude/skills/multi-agent-analysis
- Ref: 72d5025b022c77f7a51bdf5c1637c689c80e89d1
- 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/dowwie-multi-agent-analysis
- Manifest: https://skillstore.pages.dev/api/skills/dowwie-multi-agent-analysis/manifest

## Capabilities

- Classifies systems as supervisor, peer-to-peer, pipeline, market-based, or hybrid coordination models.
- Documents handoff mechanisms, context preservation, bidirectional control, and loop prevention.
- Compares shared blackboard state with message passing and per-agent isolation.
- Maps direct invocation, queue-based, and event-driven communication protocols.
- Produces a reusable Markdown analysis template for framework reviews.

## Use Cases

- Review Agent Architecture: Map the coordination model, agent inventory, handoffs, and state boundaries in an existing codebase.
- Compare Framework Designs: Evaluate how different frameworks route tasks, share state, and handle asynchronous communication.
- Plan Integration Work: Identify where orchestration, queues, events, and loop prevention need attention before implementation.

## Prompt Templates

### Identify the Coordination Model

```
Analyze this multi-agent system and classify its coordination model. Focus on control flow, agent roles, and handoff points.
```

### Map Handoffs and State

```
Review the provided agent files. Document each handoff, what context moves between agents, and where state is shared or isolated.
```

### Compare Communication Protocols

```
Compare the communication protocols in these two multi-agent frameworks. Explain latency, coupling, async behavior, and failure modes for each.
```

### Audit the Orchestration Design

```
Perform a full multi-agent analysis for this codebase. Include coordination model, agent inventory, handoff mechanics, state sharing, communication protocol, loop prevention, and recommendations.
```

## Limitations

- Does not execute, test, or validate agent framework code.
- Requires the user to provide code paths, architecture notes, or relevant files.
- Does not replace threat modeling for deployed agent systems.
- Examples are illustrative and may need production security controls.

## Best Practices

- Provide representative orchestrator, router, and communication files before asking for analysis.
- Ask for evidence locations so each coordination claim maps back to the source.
- Review loop prevention, context preservation, and state isolation together.

## Anti Patterns

- Do not ask for conclusions without sharing architecture details or code context.
- Do not treat illustrative patterns as production-ready security designs.
- Do not ignore error handling, retries, and handoff limits during analysis.

## Security Audit

- Audited at: 2026-07-06T12:20:04.415\+00:00
- Summary: All static findings were false positives caused by Markdown fences, inline code formatting, or example terminology in SKILL.md. No evidence of executable commands, credential handling, reconnaissance behavior, or prompt injection attempts was found in the reviewed skill file.

## Stats

- Views: 198
- Downloads: 9
- Favorites: 0
- Popularity score: 0