[FEATURE] Video Input Support in Claude Code
Preflight Checklist
- [x] I have searched existing requests and this feature hasn't been requested yet
- [x] This is a single feature request (not multiple features)
Problem Statement
Summary
Add the ability to paste or reference video files in Claude Code CLI, similar to the existing image/screenshot support.
Motivation
Claude Code currently supports pasting screenshots and referencing image files, which is extremely useful for UI debugging and visual context. However, there's no support for video input, which limits the tool's usefulness in scenarios like:
UI/UX debugging: Demonstrating animation bugs, transition issues, or interaction problems that can't be captured in a single frame
Screen recordings: Sharing quick recordings of bug reproduction steps
Mobile app development: Showing gesture-based interactions or scroll behaviors
Code walkthroughs: Providing video context of terminal output or build processes
Supported formats (suggested)
MP4
WebM
MOV
GIF (animated)
Implementation considerations
Video files can be large; consider adding a size limit or automatic compression
Option to extract key frames automatically if full video processing isn't feasible
Could leverage Claude's vision capabilities on extracted frames as an intermediate solution
Environment:
Platform: All (macOS, Linux, Windows)
Feature type: Enhancement
Proposed Solution
Add video input support to Claude Code with the following capabilities:
Paste video from clipboard (similar to current image paste with Ctrl+V)
Drag and drop video files into the terminal
Reference video file paths directly in prompts (e.g., @video.mp4)
Alternative Solutions
Manual frame extraction: Users can extract screenshots from videos manually, but this is cumbersome and loses temporal context
Text descriptions: Describing video content in text is imprecise and time-consuming
Priority
High - Significant impact on productivity
Feature Category
CLI commands and flags
Use Case Example
Scenario: Debugging a CSS Animation Bug
A developer is working on a React component with a complex multi-step animation. The animation works correctly on the first run but glitches on subsequent triggers.
Current workflow (without video support):
Developer: "My animation has a bug where it stutters on the second run"
Claude: "Can you describe what's happening?"
Developer: "It like... jumps? The opacity fades but then snaps back briefly"
Claude: "Could you share your CSS and the component code?"
[Multiple back-and-forth exchanges trying to pinpoint the issue]
Proposed workflow (with video support):
Developer: [pastes 3-second screen recording with Ctrl+V]
"Here's the animation bug - watch the second click"
Claude: "I can see the issue. At around 1.8 seconds, there's a flash
where the element briefly shows full opacity before fading.
This is likely caused by the animation not resetting properly.
Looking at your keyframes, you're missing animation-fill-mode: forwards
which causes it to snap back to the initial state before restarting.
Try adding:
.fade-element {
animation-fill-mode: forwards;
}"
Scenario | Video Shows | Benefit
-- | -- | --
Mobile gesture bug | Swipe interaction failing | Captures touch timing and response
Race condition | Intermittent UI flicker | Shows timing-dependent behavior
Performance issue | Scroll jank or lag | Visualizes frame drops
Integration test failure | E2E test recording | Full reproduction context
Onboarding flow review | User journey walkthrough | UX feedback with full context
</pre>
Additional Context
Related to existing image support: Users can already paste images via Ctrl+V and drag/drop with Shift held
Claude's underlying models would need to support video input for full functionality
Even partial support (e.g., automatic frame extraction with timestamps) would be valuable
16 Comments
Found 3 possible duplicate issues:
This issue will be automatically closed as a duplicate in 3 days.
🤖 Generated with Claude Code
Use FFmpeg!
Related: https://github.com/anthropics/claude-code/issues/2614
This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.
Yes please I would love this
bump! would like this, has anyone built a claude skill for this?
It would be awesome!
+1 would be really useful!
+1000
+1 this would be great
Used to having this feature in Codex, surprised to see it missing when I just testing Claude now. Really helps for troubleshooting/developing anything with a UI.
+1 would really appreciate this
Gemini can take mp4 as input but AFAIK it essentially converts the video into 1s frames and process the sequence of frames. I've been working around this with a simple tool, AI, Look! — it does screen recording with key frame pinning, so you can capture the moment, annotate it, and paste it straight into Claude Code in one click.
Works well for the animation/race condition debugging cases mentioned here where a single screenshot doesn't cut it. hope this helps
Yes, of course, that's exactly what I've done also, but it is a workaround, not a real solution.
please add this feature :)
+1 — hit this today. Use case was analysing the chain-of-thought / reasoning lines visible in a WhatsApp screen recording of another AI assistant's response. Couldn't tell Claude Code
@/path/to/video.mp4and have it parse the text on-screen frame-by-frame.Manual workaround (frame extraction via
ffmpeg -i in.mp4 -vf fps=1 out_%03d.pngthen attaching each PNG) is exactly the kind of friction the existing image/PDF support eliminated — text-on-screen in a video is the same content problem as text-on-screen in a PNG, just temporal. Even a minimal first pass (auto-extract N frames at uniform stride, treat each as an image) would unblock this category of task.