사용 방법
- 설치 명령을 복사하고 자신의 에이전트 환경에 설치하기 전에 소스 패키지를 확인하세요.
- 필요한 도구, 인증 정보, 지원 환경은 원본 문서에서 확인하세요.
- 입력과 원하는 산출물을 설명하고 출처와 결과를 대조하세요.
npx skills add https://github.com/meitu/vmake-skills --skill vmake-video-remove-watermark시작 전 확인
- 설치에는 모델 구독, API 한도, 파일 접근 권한이 포함되지 않습니다.
- 별점 수와 설치 수는 활동 지표이며 독립적인 품질·보안 검증이 아닙니다.
원본 저장소의 제목, 예시, 코드, 표, 링크, 이미지를 유지해 표시합니다.
Remove video watermarks
Capability and preparation
Use video-remove-auto for watermark removal. The service automatically identifies the content to remove and repairs the frames. Proceed within the existing authorization when the input and goal are clear. Preserve the original for comparison; check the setup reference for requests to preserve elements or control regions.
Use parameters supported by the current CLI; consult help or contracts for details. Requires vmake-labs-cli, targeting version 0.1.11, compatible with >=0.1.11 <0.2.0.
Before the first invocation, read CLI setup, resolve the executable, and check its version and tool video-remove-auto --help. If JSON fields or media limits are unclear, query the relevant vmake contracts entry. For output parsing, service questions, or interrupted execution, read runtime and recovery.
Collect input
- Obtain one actual video file or URL from the request or attachments. Do not ask again when it is clear. If several candidates are ambiguous, ask the user to choose; process multiple inputs individually only when authorized, and keep a record for each.
- Use a real readable path for a local input. Download remote originals according to CLI setup, then submit the local file. If only a media ID is available, query its session for the real URL first. If downloading fails, ask for the original local file.
- Add
-r '<session_id>'to continue a task; omit it for a new one. Follow CLI setup for defaults, authentication, and missing information. Consult current help for optional parameters and omit unspecified values. - Keep the original source, local path, and session for recovery. When the user supplies a downloaded file, continue the established goal. Do not treat failed output as a source.
Upload and submit
Use the resolved executable with the prepared local file as an individual argument:
vmake tool video-remove-auto --video-file '<local-video-path>'The CLI validates the file, uploads it through the built-in upload pipeline to obtain its CDN URL, then submits and observes the task. Do not first call create-room or add-media, or construct placeholder media references. Parameter files and separate preflight are optional; follow CLI setup for argument handling and cleanup.
Follow and deliver
- Preserve original inputs, real
session_id/room_url, and every Task ID. Read the original command's stdout and stderr separately without starting another concurrent observer. - Deliver only after
TASK_STATE_COMPLETEDand an actual usable output URL from this task. HTTP 200, stream closure, exit code 0, orend.reason=resultalone do not prove completion. Never treat an unknown state as success. - Use native previews or attachments when available, otherwise complete clickable URLs with their query parameters. Deliver every usable Part and any successful partial result. Include an actual returned session URL; do not construct one from an ID.
- For disconnection, timeout, or uncertain acceptance, query the original session and tasks first. Retry processing only after fixing a confirmed failure and obtaining authorization. If output download fails, recover the download without repeating successful processing. See runtime and recovery for details.