Argos
Visual Testing for developers
Start new thread

Media Sharing - Let AI agents put screenshots and videos on pull requests

by
•
GitHub has no API for attaching images to pull requests, so agents and CI ship blind. Argos Media Sharing fixes it: one command turns a screenshot or recording into a stable share link with ready-to-paste Markdown, and posts it on the PR automatically.

Add a comment

Replies

Best
Maker
📌
Hey Product Hunt! 👋 I'm Greg, co-founder of Argos. We built this because of a hole that kept surprising me: GitHub has no public API for comment attachments. A signed-in browser can drag a screenshot into a pull request, a coding agent working from a terminal cannot. So agents describe UI changes in prose, and reviewers merge on trust or check out the branch to see for themselves. Media Sharing closes that gap. Argos media upload turns any image or screen recording into a stable share link with ready-to-paste Markdown. Stage media on a branch while the agent works; when a pull request opens, Argos publishes it and posts one managed comment, automatically. The two details I'm proudest of: A media is an identity; every upload is a version. Re-upload after review and the URL never changes, the PR embed updates itself, and the version the reviewer commented on survives underneath. Feedback agents can act on. Reviewers pin comments to a point on the image, stored as normalized coordinates plus the exact version. An agent that can't see pixels reads the spot from the CLI, fixes it, and resolves the thread. It works from the CLI, the Node.js SDK, the REST API, and our MCP server, and there's an argos-upload skill that teaches your coding agent when a screenshot beats a paragraph: npx skills add It's live on every plan, including the free one, an image draws 1 screenshot unit from the allowance you already have, a video 25. I'd love to know: how do you get your agents' work in front of reviewers today? I'll be around all day answering questions. 🙏

That is really useful, I personally had issues with setting this up myself because of GitHub's limitations.

What do you use to get the agent to capture screenshots/recordings that are actually useful, instead of catching the app in some random state?

Maker

 nice to see you like it! On my side it runs Playwright (that is installed locally in the project for E2E + Visual Testing) and it browse the app locally, take screenshots or videos of the relevant parts and upload them. So it's not random, the agent takes care of using the good state. Also it's one-shot so we don't need the stability required for visual testing.