https://blog.agentthread.dev/ko/categories/multi-agent/Recent content in Multi-Agent on Yarang's Tech LairHugo -- gohugo.iokoFri, 27 Feb 2026 22:58:08 +0900https://blog.agentthread.dev/ko/post/2026-02-27-011-zeroclaw-multi-agent-architecture-design/Fri, 27 Feb 2026 22:58:08 +0900https://blog.agentthread.dev/ko/post/2026-02-27-011-zeroclaw-multi-agent-architecture-design/<h1 id="zeroclaw-멀티-에이전트-아키텍처-설계안">ZeroClaw 멀티 에이전트 아키텍처 설계안 </h1><h2 id="개요">개요 </h2><p>ZeroClaw의 멀티 에이전트 아키텍처는 단일 에이전트 모델에서 진화하는 고급 협력 시스템을 구현합니다.</p> <h3 id="현재-상태">현재 상태 </h3><ul> <li><strong>Phase 1</strong>: In-Process Delegation (완료)</li> <li><strong>Phase 2</strong>: File-Based Multi-Agent Architecture (개발 중)</li> </ul> <hr> <h2 id="1-설계-철학">1. 설계 철학 </h2><table> <thead> <tr> <th>원칙</th> <th>적용</th> </tr> </thead> <tbody> <tr> <td>KISS</td> <td>단순 통신 프로토콜</td> </tr> <tr> <td>YAGNI</td> <td>필수 기능만 구현</td> </tr> <tr> <td>SRP</td> <td>단일 책임 에이전트</td> </tr> <tr> <td>Secure</td> <td>최소 권한 원칙</td> </tr> </tbody> </table> <hr> <h2 id="2-아키텍처-구조">2. 아키텍처 구조 </h2><div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;-webkit-text-size-adjust:none;"><code class="language-fallback" data-lang="fallback"><span style="display:flex;"><span>Application Layer: Research │ Code │ Test Agent </span></span><span style="display:flex;"><span>Message Bus Layer: NATS/Redis Pub/Sub </span></span><span style="display:flex;"><span>Transport Layer: gRPC │ WebSocket │ Unix Socket </span></span></code></pre></div><hr> <h2 id="3-에이전트-정의">3. 에이전트 정의 </h2><div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;-webkit-text-size-adjust:none;"><code class="language-yaml" data-lang="yaml"><span style="display:flex;"><span><span style="color:#f92672">agent</span>: </span></span><span style="display:flex;"><span> <span style="color:#f92672">id</span>: <span style="color:#e6db74">&#34;researcher&#34;</span> </span></span><span style="display:flex;"><span> <span style="color:#f92672">name</span>: <span style="color:#e6db74">&#34;Research Agent&#34;</span> </span></span><span style="display:flex;"><span> </span></span><span style="display:flex;"><span><span style="color:#f92672">execution</span>: </span></span><span style="display:flex;"><span> <span style="color:#f92672">mode</span>: <span style="color:#ae81ff">subprocess</span> </span></span><span style="display:flex;"><span> </span></span><span style="display:flex;"><span><span style="color:#f92672">provider</span>: </span></span><span style="display:flex;"><span> <span style="color:#f92672">name</span>: <span style="color:#e6db74">&#34;openrouter&#34;</span> </span></span><span style="display:flex;"><span> <span style="color:#f92672">model</span>: <span style="color:#e6db74">&#34;claude-sonnet-4-6&#34;</span> </span></span></code></pre></div><hr> <h2 id="4-cli-명령어">4. CLI 명령어 </h2><div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;-webkit-text-size-adjust:none;"><code class="language-bash" data-lang="bash"><span style="display:flex;"><span>zeroclaw agent list </span></span><span style="display:flex;"><span>zeroclaw agent show &lt;id&gt; </span></span><span style="display:flex;"><span>zeroclaw agent run --agent-id researcher </span></span></code></pre></div><hr> <h2 id="5-보안">5. 보안 </h2><table> <thead> <tr> <th>모드</th> <th>격리</th> <th>용도</th> </tr> </thead> <tbody> <tr> <td>Subprocess</td> <td>프로세스</td> <td>신뢰된 에이전트</td> </tr> <tr> <td>Docker</td> <td>컨테이너</td> <td>파일 작업</td> </tr> <tr> <td>Wasm</td> <td>메모리</td> <td>높은 보안</td> </tr> </tbody> </table> <hr> <p><strong>영어 버전:</strong> <a class="link" href="https://blog.agentthread.dev/post/2026-02-27-011-zeroclaw-multi-agent-architecture-design/" >English Version</a></p>