# Agent and Chat should accept video as input

**URL:** <https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864>\
**Category:** Feature Requests\
**Created:** [June 14, 2025, 6:47pm UTC](https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864 "2025-06-14T18:47:11Z")\
**Posts on this page:** 9\
**Page:** 1

<div class="post-metadata">

**Author:** ![alexkbeck](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/alexkbeck/32/39953_2.png) [@alexkbeck](https://forum.cursor.com/u/alexkbeck)\
**Post date:** [June 14, 2025, 6:47pm UTC](https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864/1 "2025-06-14T18:47:11Z")

</div>

> Agent and Chat should accept video as input

That’s it, that’s the request.

Today I was doing some UI work and I wanted an animation effect that was hard to describe in words. I ended up doing a screen recording of a similar animation, putting that into Google AI Studio, asking it to generate code, and then giving that code to Cursor to integrate into my project.

I was honestly surprised Cursor didn’t accept video inputs.

This feature would be amazing for describing animations or complex user experience workflows to the models.

---

<div class="post-metadata">

**Author:** ![mrtornado](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/mrtornado/32/43_2.png) [@mrtornado](https://forum.cursor.com/u/mrtornado)\
**Post date:** [June 15, 2025, 2:03pm UTC](https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864/2 "2025-06-15T14:03:48Z")

</div>

yea I second this

---

<div class="post-metadata">

**Author:** ![vmath20](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/vmath20/32/78915_2.png) [@vmath20](https://forum.cursor.com/u/vmath20)\
**Post date:** [September 5, 2025, 4:39pm UTC](https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864/3 "2025-09-05T16:39:25Z")

</div>

I third this.

---

<div class="post-metadata">

**Author:** ![damianimashaun](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/damianimashaun/32/78995_2.png) [@damianimashaun](https://forum.cursor.com/u/damianimashaun)\
**Post date:** [September 7, 2025, 4:42am UTC](https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864/4 "2025-09-07T04:42:54Z")

</div>

I fourth this

---

<div class="post-metadata">

**Author:** ![user797](https://avatars.discourse-cdn.com/v4/letter/u/91b2a8/32.png) [@user797](https://forum.cursor.com/u/user797)\
**Post date:** [October 9, 2025, 2:28am UTC](https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864/5 "2025-10-09T02:28:39Z")

</div>

I fifth this

---

<div class="post-metadata">

**Author:** ![agoodkind](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/agoodkind/32/92824_2.png) [@agoodkind](https://forum.cursor.com/u/agoodkind)\
**Post date:** [January 9, 2026, 5:42pm UTC](https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864/6 "2026-01-09T17:42:51Z")

</div>

+1 on this, GPT, Gemini, etc all allow videos on their models in their webUI

---

<div class="post-metadata">

**Author:** ![Jared\_Baszler](https://avatars.discourse-cdn.com/v4/letter/j/35a633/32.png) [@Jared\_Baszler](https://forum.cursor.com/u/Jared_Baszler)\
**Post date:** [January 20, 2026, 7:19pm UTC](https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864/7 "2026-01-20T19:19:50Z")

</div>

Been a cursor user for a long while but found this today would have been most helpful trying to explain some odd UI behavior. I tried using a GIF as it suggested but that was too choppy for the agent model to pick up on it.

---

<div class="post-metadata">

**Author:** ![herringtown](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/herringtown/32/93630_2.png) [@herringtown](https://forum.cursor.com/u/herringtown)\
**Post date:** [February 3, 2026, 7:32pm UTC](https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864/8 "2026-02-03T19:32:12Z")

</div>

+11 for this ; would be so much easier than painstakingly narrating multiple screenshot and also capturing stuff that a screenshot can’t quite get (a nuanced hover state, e.g. … my current fallback for when I can’t screencap something d/t limitations of a live UX interaction that precludes using a screenshot keyboard shortcut is to literally use my iphone to capture a video then grab specific frames). Would be a gamechanger if I could just loom and narrate what I am trying to describe, pass cursor the loom link and then it could use the loom transcript + loom frame markers (e.g. thumbs-up/reactions) + whatever video it wanted to digest. Or…sure, upload a short webp/mov.

(I can appreciate the challenge of trying to support someone uploading long video and expecting quick digestion of it… ¯\_(ツ)\_/¯ )

---

<div class="post-metadata">

**Author:** ![PrinceMuichkine](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/princemuichkine/32/21061_2.png) [@PrinceMuichkine](https://forum.cursor.com/u/PrinceMuichkine)\
**Post date:** [July 14, 2026, 9:13pm UTC](https://forum.cursor.com/t/agent-and-chat-should-accept-video-as-input/103864/9 "2026-07-14T21:13:21Z")

</div>

this is such a great idea. I don’t understand why its still not available.
