OpenClaw adds subtitles and exports a video
Zhang reports an OpenClaw subtitle-and-export flow producing an MP4 with 47 Chinese subtitle segments in about ninety seconds, plus state checks that reduce duplicate-edit risk.
dev.to
Original prompt
Shared by the source author. See source notes for context.
add Chinese subtitles to my video and export as mp4
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Wei Zhang
- Original account
- Read the source ↗
- Published
- 18 Mar 2026
- Source link checked
- 15 Oct 2026
- Reported result
- Zhang reports an OpenClaw subtitle-and-export flow producing an MP4 with 47 Chinese subtitle segments in about ninety seconds, plus state checks that reduce duplicate-edit risk.
- Human involvement
- The user supplies the desired edit request. The source does not document an actual subtitle-quality inspection or independent evaluation of the exported media.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Send the Chinese subtitle instruction to NemoVideo through its streaming API.
- Interpret the backend result and queue export after the edit completes.
- Call the rendering API directly, poll completion and download the finished MP4.
Wei Zhang built an OpenClaw skill around the NemoVideo backend so people could edit video through chat. His account describes a concrete request to add Chinese subtitles and export an MP4. The agent routes the editing instruction to the backend, interprets its response and then handles rendering as a separate API operation.
That separation fixed a costly failure. Forwarding an export request to the generation backend could create new content and charge credits rather than render the current edit. The skill instead calls the render endpoint directly, polls for completion and sends the downloaded file. The described subtitle run returned 47 segments and a finished MP4 in roughly ninety seconds.
Zhang also documents silent edits. Some backend operations updated the timeline without returning text, so retrying could duplicate changes and charges. The revised skill compares state before and after the request and reports the observed difference. These are builder-reported outcomes and timings. The account does not establish transcription quality or correctness for every language. The user specifies the desired edit; the source does not document a separate quality inspection of the exported media.
Prompt
Original prompt
Source: Blog · Wei Zhang
Frequently asked questions
Why separate export from editing?
Sending export through the generation backend could start another charged generation instead of rendering the existing timeline.
How are silent edits handled?
The skill compares state before and after the request and reports what actually changed.
What output does the example return?
A downloaded MP4 with burned-in Chinese subtitles, reported as 47 segments.
More details
Remaining steps
- For silent edits, compare timeline state rather than blindly retrying.
About this account
Send the Chinese subtitle instruction to NemoVideo through its streaming API. Interpret the backend result and queue export after the edit completes.
Firsthand account read. Primary DEV Article datePublished 2026-03-18T06:22:15Z; March 22 edit is not substituted. Builder-reported run and timing; output quality not independently evaluated. Reported outcomes were not independently reproduced. Original prompt box contains a short excerpt of the author's published instructions.
Credit: Wei Zhang
Source published 18 Mar 2026
Source checked 15 Oct 2026
Tools and topics
OpenClaw · NemoVideo API · Server-Sent Events · Render API
OpenClaw · NemoVideo API · Server-Sent Events
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.