d3k

작성자: vercel

독립형 AI 앱(Codex, Cursor, Claude Code)에서 d3k 부트스트랩: dev3000 감지/설치, d3k를 런타임으로 시작, 통합 로그 및 CDP 브라우저 사용…

npx skills add https://github.com/vercel-labs/dev3000 --skill d3k

d3k Agent Runtime

d3k is the local web runtime for this task. It starts the dev server behind a stable Portless URL, owns a project-stable Chrome profile, and records server logs, browser console output, network activity, interactions, and screenshots in one timeline.

When this skill triggers, operate d3k. Do not merely tell the user how to run it.

Interpret the Request

  • "Let me test/dev my project with d3k": prepare the runtime and headed browser, confirm it is ready, then hand control to the user. Wait for them to reproduce the issue before inspecting evidence.
  • "Test/debug/fix this with d3k": prepare the runtime, then drive the managed browser and investigate autonomously.
  • If ambiguous, start the runtime and browser first. That action is safe and useful for either path.

Start or Reuse d3k

Run from the project root.

  1. Check canonical Portless readiness before starting any app process:
d3k portless status --json

If "setupRequired": true, run:

d3k portless setup

On macOS this opens the system administrator authorization dialog. Wait for the user to approve it, then rerun d3k portless status --json. Do not start d3k until "canonical": true, "serviceInstalled": true, and "setupRequired": false. If authorization is declined or unavailable, stop and explain the blocker. Never silently substitute direct localhost.

  1. Check for an existing project runtime:
d3k status --json

Reuse it only if it reports "running": true, "ready": true, "routing": "portless", "browserConnected": true, and a genuinely port-free https://...localhost appUrl. If an active session reports "routing": "direct", stop that retained d3k session and restart after Portless is ready. Do not start a second dev server or browser.

  1. If d3k is not installed, install it:
bun install -g dev3000

Use npm install -g dev3000 only when Bun is unavailable.

  1. Start d3k with the agent's shell/process tool as a retained background or yielded session (non-TUI by default):
d3k -t

Do not wait for this long-running command to exit. Keep its process/session handle so you can monitor or stop it later. Prefer the execution tool's background/session support over shelling with &.

If the target URL is already known, pass it so the managed browser opens there:

d3k -t --app-url "<url>"

Let d3k auto-detect the package manager, dev command, and port. Add --command, --script, or --port only when detection is wrong or the user specified them.

  1. Poll until the runtime is ready:
d3k status --json

A successful status response is the readiness boundary only when it reports "ready": true, "routing": "portless", "browserConnected": true, and a port-free https://...localhost appUrl. The underlying app port may change between runs. If startup fails, inspect the retained process output and d3k logs --type server; do not launch a separate dev server.

User-Driven Testing

When the user says "let me test":

  1. Confirm the status response includes the app URL and "browserConnected": true.
  2. Tell the user the monitored browser is ready.
  3. Keep the d3k process running and wait for the user to reproduce the behavior.
  4. When they report that it happened, begin with:
d3k errors --context
d3k logs -n 200

Do not replace the headed browser with automation while the user is testing.

Agent-Driven Testing

Drive the exact browser d3k is monitoring:

d3k agent-browser snapshot -i
d3k agent-browser click @e2
d3k agent-browser fill @e3 "text"
d3k errors --context

Use --require-d3k-browser when opening a URL so failure cannot silently create another browser:

d3k agent-browser --require-d3k-browser open "<url>"

After every reproduction or code change, replay the relevant interaction and check d3k errors --context again.

Evidence Commands

Prefer these over ad-hoc log scraping:

d3k status --json
d3k errors --context
d3k logs -n 200
d3k logs --type browser
d3k logs --type server

Artifacts live under ~/.d3k/<project>/, including session.json, logs, screenshots, and the Chrome profile.

Browser and Auth Safety

d3k must own browser startup by default. Its per-project Chrome profile preserves login state, cookies, and local storage.

For Google OAuth, Supabase auth, and other auth-sensitive flows, never substitute raw Chrome, Playwright, a browser MCP session, manual CDP attachment, or agent-browser --profile. Those paths use a different browser/profile and can trigger "This browser or app may not be secure."

If the managed browser is unavailable, stop or interrupt the retained d3k process and restart d3k cleanly. Do not work around it by creating another browser.

Use --headless only for CI or when explicitly requested. Use --servers-only only when browser monitoring is intentionally unwanted.

Operating Rules

  • Do not run npm run dev, bun run dev, or another dev server alongside d3k.
  • Do not start a second d3k when d3k status --json reports an active one.
  • Keep d3k alive across edits and retests.
  • Preserve the project-stable Chrome profile unless the user explicitly asks for a fresh profile.
  • Leave the runtime running when handing a headed browser to the user; stop it only when asked or when the task requires a clean restart.
  • Canonical Portless HTTPS routing is required by default. Never add --no-portless or set PORTLESS=0 unless the user explicitly requests direct localhost routing.

vercel의 다른 스킬

vercel
vercel
로컬 개발 및 테스트를 위한 Vercel REST API 에뮬레이션입니다. 사용자가 로컬에서 Vercel API 엔드포인트와 상호작용하거나 Vercel 통합을 테스트해야 할 때 사용합니다.
cron-jobs
vercel
Vercel Cron Jobs 구성 및 모범 사례. vercel.json에서 예약된 작업을 추가, 편집 또는 디버깅할 때 사용합니다.
codegen
vercel
json-render을 위한 코드 생성 유틸리티입니다. UI 명세서에서 코드를 생성하거나, 사용자 정의 코드 내보내기를 구축하거나, 명세서를 탐색하거나, props를 직렬화할 때 사용합니다.
next-best-practice
vercel
Next.js 모범 사례 - 파일 규칙, RSC 경계, 데이터 패턴, 비동기 API, 메타데이터, 오류 처리, 라우트 핸들러, 이미지/폰트 최적화,…
benchmark-sandbox
vercel
Vercel Sandbox에서 vercel-plugin eval 시나리오를 로컬 WezTerm 패널 대신 실행합니다. Claude Code와 플러그인이 사전 설치된 임시 마이크로VM을 프로비저닝합니다.
write-guide
vercel
점진적인 예제를 통해 실제 사용 사례를 가르치는 기술 가이드를 제작합니다. 개념은 독자가 필요로 할 때만 소개됩니다.
benchmark-testing
vercel
벤치마크 테스트 프로젝트를 생성하고 실행하여 실제 시나리오에서 vercel-plugin 스킬 인젝션을 테스트합니다. 격리된 디렉토리를 설정하고, 설치하며…
ai-gateway
vercel
Vercel AI Gateway 전문가 안내. 모델 라우팅, 제공업체 장애 조치, 비용 추적 또는 통합된 방식을 통해 여러 AI 제공업체를 관리할 때 사용합니다.