跳至正文
OpenAgentSkill
浏览已开放 · 目录记录日期 2026-11-21 · 自动更新和账户服务暂时关闭。
视频与音频

turn-long-video-into-shorts

Turn a long recording, podcast, interview or webinar into individually reviewed short MP4s with Valmera. Use when the user wants several self-contained highlights with framing and captions adapted to the intended platform.

ABO3SKRALMASAOODI · 已有静态检查记录

安装与使用

运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。

npx skills add ABO3SKRALMASAOODI/valmera-mcp --skill turn-long-video-into-shorts

技能说明


name: turn-long-video-into-shorts description: Turn a long recording, podcast, interview or webinar into individually reviewed short MP4s with Valmera. Use when the user wants several self-contained highlights with framing and captions adapted to the intended platform. license: MIT

Turn a long video into shorts with Valmera

Each short has its own project, edit version and final export. Select complete stories from the source, then review the framing and captions on each deliverable.

Connection and cost

Connect to https://valmera.io/mcp/server and authenticate using the client setup guide. Discover the current tools and input schemas in the connected session; the names below describe a workflow, not a permanent catalog.

Account creation and uploads are free; editing requires a paid Valmera plan. Before billable work, explain the intended scope and use any cost estimate exposed by the account. If an exact quote is unavailable, say so and obtain approval for a bounded preview before proceeding. Reuse an existing explicit scope and budget approval; ask again if the work would exceed it. Do not buy credits or upgrade a plan without authorization. Review a preview before an approved final export.

Clips from a multi-hour recording

Check the indexed coverage before searching a late passage. Studio indexes the first three hours of a main recording. For a passage outside that window, use a section containing the full question and answer. Keep the original time range and the section’s actual start time in the brief. A passage at 03:20:00 in the original starts at 00:30:00 in an excerpt beginning at 02:50:00. Select the story on the excerpt’s clock and use the original clock when telling the creator where it came from.

The long-video checker reads current upload limits and can convert an original SRT into captions for the section. Rebase source captions once. Captions generated from the excerpt already use its clock. Check the first and last cues against the excerpt and review timing again after cuts.

Workflow

  1. Set up the source. Open the intended project, or use create_project with the current shorts-project schema and upload the recording. Wait for index_status to report ready. Reuse the user's requested count, duration, aspect ratio and audience.
  2. Read the conversation and visual context. Use the transcript plus available editorial and frame evidence. Choose moments that make sense without the rest of the recording, preserve the speaker's meaning and finish their thought. A hook is useful only if the rest of the clip delivers it.
  3. Resolve the selection. If the user has chosen moments, use them. Otherwise present candidate source ranges, a working title and a short reason when selection needs their input; proceed within an already approved selection brief.
  4. Create and track the shorts. Use the current make_shorts schema for the selected stories, follow shorts_status, and open each result with open_short. Follow existing jobs rather than resubmitting an operation that is still running.
  5. Polish each story. Apply the requested framing, captions, opening text and sound. Check two-person scenes and camera changes rather than assuming one crop suits the whole clip. Use apply_short_edit_batches when its current schema fits an approved batch; validate a representative preview before expanding the treatment.
  6. Review each deliverable. Render a preview, wait for its job, and inspect the opening, cut boundaries, captions and framing. Distinguish source times from output times in revision notes. Identify each short's project and reviewed edit version.
  7. Export approved versions. For each authorized final, call export_final with that short's project and reviewed edl_version, then wait_for_job, then download_url with kind="final". Report the actual files and any unfinished shorts separately. Do not present a preview or job receipt as a final export.

Delivery and outcome

Confirm caption language and output files against current project options. State the destination and review each aspect ratio using the delivery guide. Publishing to an account is a separate action and needs the user's authorization.

No tool can guarantee virality. Explain why the chosen stories fit the brief and which parts of the result were actually reviewed. For one finished edit rather than several stories, use edit-video-with-valmera when installed.