# An open-source tool that turns client feedback videos into Cursor implementation prompts

**URL:** <https://forum.cursor.com/t/an-open-source-tool-that-turns-client-feedback-videos-into-cursor-implementation-prompts/170498>\
**Category:** Built for Cursor\
**Tags:** context\
**Created:** [September 3, 2026, 7:41pm UTC](https://forum.cursor.com/t/an-open-source-tool-that-turns-client-feedback-videos-into-cursor-implementation-prompts/170498 "2026-09-03T19:41:45Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Zulelee](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/zulelee/32/77139_2.png) [@Zulelee](https://forum.cursor.com/u/Zulelee)\
**Post date:** [September 3, 2026, 7:41pm UTC](https://forum.cursor.com/t/an-open-source-tool-that-turns-client-feedback-videos-into-cursor-implementation-prompts/170498/1 "2026-09-03T19:41:45Z")

</div>

Two weeks ago, a client sent us feedback on a tool we built for them, in eight separate screen recordings, each five to eight minutes long.

I watched every recording and took notes, but I still worried that I might miss a small yet important detail during implementation. When feedback is spread across nearly an hour of video, it’s surprisingly easy to lose context.

So I built **Video2Prompt**.

It transcribes each recording, analyzes what the client is saying alongside what they’re showing on screen, and produces a structured implementation prompt that can be used directly in Cursor.

Instead of giving Cursor something vague like:

> Add a Customers tab to the dashboard.

It captures the surrounding context: what should appear in the tab, how it should behave, what part of the interface the client was referring to, and any other requirements mentioned during the recording.

This doesn’t replace watching the feedback. I still think it’s important to hear the client explain the problem in their own words. It simply makes sure nothing gets lost between watching the recording and implementing the changes.

You can also provide the project path so the generated prompt maps the feedback to relevant files and symbols in the codebase.

```auto
video2prompt analyze input/client-feedback.mp4 \
  --project /path/to/project \
  --provider cursor

```

The tool is open source and also includes an extraction mode for turning videos into reusable content playbooks.

GitHub: [GitHub - Zulelee/Video2Prompt: Turn videos into actionable prompts for your AI tools. · GitHub](https://github.com/Zulelee/Video2Prompt)

I’d love to hear what other Cursor users think, and whether video-based feedback is also a regular part of your workflow.

 ![Screenshot 2026-09-04 at 12.40.06 AM](https://us1.discourse-cdn.com/cursor1/original/3X/9/9/995a0bc186fed4dc9c04d299f5f6b440e7cc37f7.png)