Turning Instagram Explore into a Spotify playlist

Explore is where I hear music I wouldn't find anywhere else. But a post plays for a few seconds, I scroll on, and finding the song again means digging back for the post and searching Spotify by hand.

I wanted the list to build itself while I scroll: nothing to tap, nothing to copy.

I set the rules before writing anything. No server, no account, no analytics. The list stays in the browser, and the only service it talks to is Spotify, through the person's own account.

No automation either. You scroll Explore yourself; the extension never scrolls, clicks, likes, follows or posts. It makes no requests of its own to Instagram and never asks for an Instagram login.

That ruled out the easy routes, and it shaped everything that followed.

Instagram's web app already fetches post data as you scroll, and that data names the song. The extension reads those responses in the page, keeps the songs from Explore, and links each one back to its post.

The details took the most iteration. The first screen of Explore arrives inside the page itself, not as a request, so early builds missed it. Explore also loads parts of the home feed, which had to be kept out. Each fix came from my own sessions and a debug export I built into the panel.

Each song is matched on Spotify, preferring the original over remixes, live versions and covers. Matching only on whole words stopped it picking "Song 1" for "Song 10".

Songs play right in the panel with Spotify's own player, or in a small window that stays on top. Untick the ones you don't want and send the rest to a playlist in one click, or turn on auto-add and new finds go straight in. When Spotify doesn't have a song, the panel offers a YouTube search, just to hear it.

The prototype tried to show everything it heard: tabs for licensed songs and creators' original audio, a checkbox to limit it to Explore, and a count of how many posts each song turned up in. The tabs and the checkbox turned out to be noise. The rebuild kept songs from Explore only and dropped both; the post count stayed, shown only when a song turns up more than once.

That left one long list with the playlist controls at the bottom, and the Spotify connection in a fold-out at the top. Using it every day showed the next problem: the thing you act on was the furthest away. So the playlist moved to the top and stays pinned while the list scrolls, and the Spotify connection moved to its own Settings screen.

Then the playlist block was reordered around the main action, Add first, and the rows started saying what the button would do: a "New" mark on songs it would add, a green check on songs already in the playlist, and select all or none above the list.

With the layout settled, I explored four looks on one canvas, each with the same songs and controls so only the styling changed: Night Deck, Daylight, Plum and Coral, and Mono.

This round was about type, colour and shape: corner radii, how the player sits, and whether each song is a row or its own card.

I picked Daylight: a cool grey ground, white cards, a cobalt accent and Manrope. It changed shapes as well as colours. The player and playlist sit in a white card, and each song is its own bordered card.

The navy look stayed on as the dark theme. Settings offers Light, Dark, or follow the system, which is the default.

Spotify only lets a developer app have five users, so a shared app couldn't work for anyone else. Each person makes their own free Spotify app instead.

That's a developer task, so I turned it into a short guided walkthrough. The extension fills in Spotify's form for you and reads back the details it needs. It never accepts Spotify's terms or saves the form on your behalf.

The code is public, with a privacy policy that says exactly what's stored and where. A store build is submitted and waiting on review.

I designed it, built it with AI, wrote the tests that drive a real Chrome, and wrote the listing. It's the kind of product loop I like to stay close to: notice a small frustration, set the constraints, ship something real.