# Design Reliable Multi-Agent Systems

Multi-agent designs often add cost and coordination failures without improving outcomes. This skill helps select topologies, handoffs, consensus methods, and recovery controls.

## Install

```bash
npx skillstore add muratcankoylan/multi-agent-patterns
```

## Metadata

- Status: approved
- Slug: muratcankoylan-multi-agent-patterns
- Skillstore revision: r2
- Version status: missing
- Tree hash: 4eec11db3348db47d2716f89b068398a635296fe7de1f0e36403b0d1269eea6c
- Author: muratcankoylan
- GitHub username: muratcankoylan
- License: MIT
- Repository: https://github.com/muratcankoylan/Agent-Skills-for-Context-Engineering/tree/main/skills/multi-agent-patterns
- Ref: 02be9409c79ca1183f7844009c14d9df684d0cf9
- 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: filesystem, external\_commands, network
- Quality score: 80
- Quality tier: silver
- Public page: https://skillstore.pages.dev/skills/muratcankoylan-multi-agent-patterns
- Manifest: https://skillstore.pages.dev/api/skills/muratcankoylan-multi-agent-patterns/manifest

## Capabilities

- Compares supervisor, peer-to-peer, swarm, and hierarchical architectures.
- Explains context isolation through delegation, instructions, and shared files.
- Defines explicit handoff, routing, weighted voting, and debate patterns.
- Provides Python coordination utilities for messaging, supervision, consensus, and failure handling.
- Documents framework patterns for LangGraph, AutoGen, and CrewAI.

## Use Cases

- Plan a research workflow: Design parallel research, analysis, verification, and writing roles with clear aggregation rules.
- Improve agent handoffs: Define state transfer, acknowledgments, validation, and timeout behavior for customer service routing.
- Review architecture costs: Compare a proposed multi-agent topology with a single-agent baseline before implementation.

## Prompt Templates

### Choose an architecture

```
Choose a multi-agent pattern for [task]. Explain why multiple agents are justified, required roles, handoffs, and main trade-offs.
```

### Design a supervisor workflow

```
Design a supervisor workflow for [goal] with three workers. Define routing, worker inputs, outputs, aggregation, and failure handling.
```

### Create a swarm protocol

```
Create a peer-to-peer handoff protocol for [domain]. Specify transfer conditions, state fields, acknowledgments, timeouts, and convergence checks.
```

### Audit a production topology

```
Audit this multi-agent design: [design]. Compare costs, context boundaries, bottlenecks, error propagation, consensus quality, retries, and single-agent alternatives.
```

## Limitations

- The included supervisor workflow simulates worker responses instead of calling real agents.
- Framework examples require external packages and application-specific integration.
- The skill does not provision hosted agents, sandboxes, or persistent infrastructure.
- Performance and token-cost claims require validation against each workload.

## Best Practices

- Start with the fewest agents needed for genuine context isolation.
- Define typed handoffs, validation rules, timeouts, and ownership boundaries.
- Benchmark quality, latency, and token cost against a single-agent baseline.

## Anti Patterns

- Do not create agents only to imitate organizational job titles.
- Do not let supervisors repeatedly paraphrase complete worker responses.
- Do not use unweighted majority voting for critical factual decisions.

## Security Audit

- Audited at: 2026-08-09T09:53:05.828\+00:00
- Summary: All 46 static findings are false positives caused by documentation syntax, ordinary Python identifiers, illustrative filesystem examples, or trusted documentation links. No command execution, credential exposure, reconnaissance, prompt injection, or malicious network behavior was found.

## Stats

- Views: 265
- Downloads: 17
- Favorites: 0
- Popularity score: 0