Claude Code: Now in Beta in Zed
TL;DR Highlight
Zed editor now runs Claude Code natively inside the editor via its own open standard ACP — not a one-off integration, but a universal protocol for any AI agent.
Who Should Read
Developers who frequently use Claude Code but find terminal switching cumbersome, or those waiting for Claude Code support to switch from Cursor to Zed.
Core Mechanics
- Zed created ACP (Agent Client Protocol), an open standard so any AI agent that implements it can connect to Zed (and any ACP-compatible editor) — not locked to a specific agent.
- Claude Code can now run natively within Zed's Agent Panel. Changes across multi-file edits can be reviewed in real-time and accepted/rejected at the code hunk level.
- ACP wraps the Claude Code SDK via an adapter. Claude Code runs as an independent process, and the adapter translates its actions into ACP's JSON RPC format for the Zed UI.
- The Claude Code adapter source is open-sourced under Apache license. Neovim's CodeCompanion plugin has already adopted ACP, enabling the same Claude Code experience in Neovim.
- Gemini CLI is also integrated via ACP. Users can use Claude Code, Gemini CLI, and other ACP-compatible agents side by side in the Agent Panel.
- GitHub user Xuanwo built an ACP implementation for Claude Code right after launch, which the Zed team publicly credited. A case of community contribution accelerating protocol adoption.
Evidence
- First comments noted it 'feels rushed.' /compact is unsupported so 'Prompt too long' errors with no workaround, Plan mode switching doesn't work, and model switching is impossible — critical limitations for actual use.
- Multiple comments pointed out missing features vs CLI: no Plan Mode, no /clear or /new, no model switching, auto-accept not actually auto-accepting. Some said using CLI inside Zed's terminal is better at this point.
- SSH remote development doesn't work, making it impractical for developers who primarily use remote environments.
- The ACP standardization effort was viewed positively by some: 'Like TLS, OAuth 2.0, and ECMAScript, multiple players building a common standard matures the ecosystem' — ACP is the right long-term direction vs proprietary integrations.
- Anthropic Max subscribers ($200/month) using Opus 4.1 raised concerns about API key costs being too high, questioning whether the Zed integration provides a practical path for these subscription users.
How to Apply
- If you use Claude Code and want visual confirmation of multi-file refactoring changes, update to the latest Zed and select Claude Code from the Agent Panel's Plus menu. Caveat: Plan Mode and /compact are missing, so test with simple tasks first.
- If you primarily use Neovim and need Claude Code editor integration, CodeCompanion plugin already adopted ACP — Claude Code works in Neovim without extra setup.
- If developing your own AI agent/tool with editor integration in mind, reference the ACP spec and open-source Claude Code adapter code to make your agent ACP-compatible for automatic connection to Zed, Neovim, and other ACP-adopting editors.
Terminology
Related Papers
Show HN: OpenKnowledge – open source AI-first alternative to Obsidian/Notion
Git 기반 동기화와 Claude/Codex/Cursor 연동을 내장한 로컬 우선 마크다운 에디터로, AI 에이전트의 두 번째 뇌(LLM Wiki)로 활용할 수 있는 오픈소스 도구다.
The Unfireable Safety Kernel: Execution-Time AI Alignment for AI Agents and Other Escapable AI Systems
AI 에이전트가 자신의 안전장치를 우회할 수 없도록, 에이전트 프로세스 바깥에 수학적으로 증명된 강제 통제 게이트를 배치하는 아키텍처
RubyLLM: A Ruby framework for all major AI providers
OpenAI, Claude, Gemini 등 주요 AI 프로바이더를 단일 인터페이스로 통합한 Ruby 프레임워크로, Rails 통합과 에이전트 기능까지 지원해 Ruby 개발자가 AI 기능을 빠르게 붙일 수 있다.
Qwen-AgentWorld: Language World Models for General Agents
Alibaba Qwen 팀이 AI 에이전트가 행동 결과를 미리 시뮬레이션할 수 있는 'Language World Model'을 공개했다. 에이전트 훈련과 실행 경로 검증에 새로운 패러다임을 제시하는 연구다.
SHERLOC: Structured Diagnostic Localization for Code Repair Agents
버그 위치만 알려주는 게 아니라 '왜, 어떻게 고쳐야 하는지'까지 진단 리포트를 생성해서 코드 수정 에이전트의 성능을 높이는 training-free 프레임워크
Show HN: peerd – AI agent harness that runs entirely in your browser
백엔드 서버 없이 Chrome/Firefox 확장 프로그램으로만 동작하는 AI 에이전트 실행 환경으로, 브라우저 탭을 직접 조작하고 WASM Linux VM까지 구동할 수 있어 프라이버시와 보안을 동시에 챙길 수 있다.