스킬 agent-desktop
📦

agent-desktop

콘텐츠 리비전 r1 중간 위험 ⚙️ 외부 명령어🌐 네트워크 접근📁 파일 시스템 접근

Automate Desktop Apps With Accessibility Controls

Desktop workflows often require repeated navigation, data entry, and checks across application windows. This skill guides agents through scoped accessibility actions and verification using agent-desktop.

지원: Claude Codex Code(CC)
Maintain this Skill? Claim it 📊 67 적절함

내 에이전트로 설치

이 요청을 에이전트에 복사하세요. 공식 Skill 페이지와 매니페스트가 포함되어 있습니다.

에이전트 요청
Review the Skillstore skill "agent-desktop" from https://skillstore.io/skills/vercel-labs-agent-desktop.md and its manifest at https://skillstore.io/api/skills/vercel-labs-agent-desktop/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.

에이전트는 계속해서 계획을 표시하고 보안 정책에서 요구하는 확인을 요청해야 합니다.

에이전트가 읽기 쉬운 리소스

AI 에이전트, 크롤러, 스크립트가 전체 페이지 대신 정리된 컨텍스트만 필요할 때 이 링크를 사용하세요.

Markdown 상세 GET /skills/vercel-labs-agent-desktop.md 서명된 manifest GET /api/skills/vercel-labs-agent-desktop/manifest 서명된 lockfile GET /api/skills/vercel-labs-agent-desktop/lockfile

테스트해 보기

"agent-desktop" 사용 중입니다. Inspect the active Notes window without changing it.

예상 결과:

Example report: The selected Notes window exposes a sidebar, a note list, and an editor. No controls were activated.

"agent-desktop" 사용 중입니다. Fill the contact form with approved details and stop before submission.

예상 결과:

  • Example report: Name and email fields match the approved values.
  • The selected option was checked after entry.
  • The form remains open and was not submitted.

"agent-desktop" 사용 중입니다. A Save shortcut was delivered, but its wait timed out. Check before retrying.

예상 결과:

Example report: The save dialog is already open. The shortcut was not repeated. Confirmation is pending in the dialog.

보안 감사

중간 위험
v1 • 2026. 10. 31. 버전별 보고서 열기

All 67 static findings are benign documentation matches, including Markdown backticks, keystroke delivery, local session storage, and loopback CDP endpoints. One semantic finding concerns pointer dispatch after inconclusive hit tests, which can misdirect input despite visibility checks. No evidence found of credential theft, covert keylogging, external exfiltration, or audit-directed prompt injection in the five reviewed files.

5
스캔된 파일
846
분석된 줄 수
0
검토 항목
0
오탐 무시됨

확인된 보안 우려 사항 (1)

중간
references/commands-interaction.md:71-74
Pointer Delivery Proceeds After Inconclusive Hit Tests
The documentation states that receives_events fails open when a hit test is inconclusive. Pointer input can reach an obscuring control or window, causing unintended actions or disclosure of typed data.
The reference explicitly states that unknown hit-test results still dispatch input. The implementation was not supplied, so this finding concerns documented behavior.

위험 요인

⚙️ 외부 명령어 (33)
references/commands-system.md:171 references/workflows.md:187 references/workflows.md:188 SKILL.md:20 SKILL.md:24 SKILL.md:28 SKILL.md:29 SKILL.md:30 SKILL.md:34 SKILL.md:36-41 SKILL.md:41-43 SKILL.md:43-44 SKILL.md:44 SKILL.md:45-46 SKILL.md:46 SKILL.md:50 SKILL.md:53-54 SKILL.md:54 SKILL.md:55-59 SKILL.md:59 SKILL.md:61 SKILL.md:62-63 SKILL.md:63 SKILL.md:64 SKILL.md:65-69 SKILL.md:69 SKILL.md:73-77 SKILL.md:77 SKILL.md:78-79 SKILL.md:79 SKILL.md:24 SKILL.md:24 SKILL.md:24
🌐 네트워크 접근 (3)
references/commands-system.md:33 references/commands-system.md:33 references/commands-system.md:34
📁 파일 시스템 접근 (8)
references/commands-system.md:110 references/commands-system.md:168 references/commands-system.md:110 references/commands-system.md:168 references/commands-system.md:10 references/commands-system.md:102 references/commands-system.md:104 references/commands-system.md:105
감사자: codex
이 보고서 공유 및 인용

버전이 지정된 평가 보고서, 중립 배지, 임베드 카드 및 인용을 공유하세요. Skillstore는 근거를 보고하지만 이 Skill의 안전 여부를 판단하지 않습니다.

버전별 보고서 열기
보안 평가

보고서 링크 복사

https://skillstore.io/skills/vercel-labs-agent-desktop/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown 배지

[![Skillstore security assessment](https://skillstore.io/badges/skills/vercel-labs-agent-desktop/security.svg)](https://skillstore.io/skills/vercel-labs-agent-desktop?utm_source=security_passport_badge)

HTML 배지

<a href="https://skillstore.io/skills/vercel-labs-agent-desktop?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/vercel-labs-agent-desktop/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

임베드 카드

<iframe src="https://skillstore.io/embed/skills/vercel-labs-agent-desktop.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>
학술 인용 (APA · BibTeX · CFF)

APA 인용

vercel-labs. (2026). agent-desktop security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/vercel-labs-agent-desktop/audits/1

BibTeX 인용

@techreport{vercel-labs-vercel-labs-agent-desktop-2026, author = {vercel-labs}, title = {agent-desktop security audit report (audit version 1)}, institution = {Skillstore}, year = {2026}, number = {1}, url = {https://skillstore.io/skills/vercel-labs-agent-desktop/audits/1}, note = {Author version unspecified} }

CITATION.cff

cff-version: 1.2.0 message: "If you use this Skill, cite its author and this versioned security audit report." title: "agent-desktop security audit report (audit version 1)" version: "unspecified" type: report authors: - name: "vercel-labs" date-released: "2026-10-31" url: "https://skillstore.io/skills/vercel-labs-agent-desktop/audits/1" identifiers: - type: other value: "skillstore:vercel-labs-agent-desktop:audit:1" description: "Skillstore immutable audit report identifier"

Skillstore 점수

이 점수의 이유 증거 신뢰도: 중간
45
아키텍처
85
유지보수성
87
콘텐츠
65
커뮤니티
83
사양 준수

만들 수 있는 것

Complete Repeated Desktop Forms

Enter approved data into application fields, select options, and verify the completed form before submission.

Check Desktop UI Workflows

Inspect controls, exercise menus and dialogs, and distinguish verified changes from uncertain input delivery.

Review Application Notifications

List notifications from a selected app and act on specific entries only after checking their current fingerprints.

이 프롬프트를 사용해 보세요

Inspect an Application 
Check agent-desktop status and permissions, then inspect [app]. List its visible controls without changing anything.
Fill a Form Without Submitting 
In [app], fill [fields] with [approved values]. Use current references and verify each field. Stop before submitting or enabling physical input.
Navigate a Save Dialog 
In [app], save the current document to [approved destination]. Inspect each menu or dialog surface. Ask before overwriting files or using physical input.
Run a Scoped Multi-Step Workflow 
Run [approved workflow] in [app] within a dedicated session. Use event waits and stop batches on errors. Verify uncertain deliveries before any repeat. Keep screenshot tracing off.

모범 사례

  • Check status and permissions, then scope reads and actions to the intended app, window, or overlay.
  • Use fresh qualified references, event waits, and independent state checks before repeating an uncertain action.
  • Require explicit permission for physical input and minimize sensitive artifacts, clipboard reads, and CDP exposure.

피하기

  • Retrying a click, submission, or toggle merely because a wait timed out.
  • Switching to physical input after a headless refusal without explicit user permission.
  • Sharing unredacted screenshots, notifications, or clipboard contents, or running parallel writers against one editor.

자주 묻는 질문

Which operating systems are supported?
The guide covers macOS, Windows, and Linux. Available surfaces, permissions, and action mechanisms vary by platform.
Does this skill install the desktop tool?
No. It requires an available agent-desktop binary. Check status and load the platform reference before starting.
Can it work without taking keyboard or mouse focus?
Supported semantic actions are headless by default. Physical gestures require headed mode and explicit user permission; some actions vary by operating system.
Does success mean the application saved my changes?
Not necessarily. Field verification proves the observed field value, not document acceptance. Check the application's resulting state independently.
Are captured records always redacted?
No. Trace lines redact sensitive fields, but optional screenshot artifacts are unredacted. Notification output is verbatim, and clipboard reads may expose private data.
Is a browser automation client required?
No. Accessibility is the default path. An optional CDP client can control freshly launched Chromium apps; close the debug-enabled app afterward.

개발자 세부 정보

작성자

vercel-labs

라이선스

MIT

Skillstore 리비전

r1

버전 안내

작성자가 버전을 선언하지 않았습니다.

리포지토리

https://github.com/vercel-labs/agent-desktop/tree/5261b5287b5208e7c4bf3713df6f66faaaf2c4cc/skills/agent-desktop

참조

443e0d99cb0af7754aabaecc5bf8b6dcf2dae44d

유지보수 최신성

2026. 10. 31.

사용량

0 다운로드 · 0 조회

파일 구조

📁 references/

📄 commands-interaction.md

📄 commands-observation.md

📄 commands-system.md

📄 workflows.md

📄 SKILL.md

vercel-labs의 다른 스킬

전체 보기

관련 스킬

전체 보기