Can a Shortcut Turn an Instagram Reel Link into a Transcript?

With XAPN today, a public Reel link can become text in the browser, but not through a fully automatic link-in/text-out shortcut. You still paste the link into the tool, select Get transcript, wait for the result, and copy it. The site's published API validates a Reel URL; it does not return the spoken words.

Open the browser tool
A public Reel link reaches XAPN's browser tool, where audio retrieval and speech recognition produce copyable text. The public guide API only checks the URL shape, so it cannot return a transcript to a shortcut.
The browser tool and the read-only guide API solve different steps.

What worked in our test

On October 6, 2026, we opened the live XAPN homepage in the Codex in-app browser. Its public example Reel link was already in the input. Selecting Get transcript first showed Fetching the Reel audio locally.... The button was disabled while the request ran. The same page later showed Done. and a nonempty English result. We did not copy the creator's speech into this article, and we did not measure an end-to-end time.

The test also exposed a limit worth keeping in view: a separate request to the live audio endpoint for that example returned no bytes before a 20-second client timeout, while the browser request eventually completed. These were different requests, not a controlled reliability comparison. One browser success does not prove every automated fetch will finish on schedule.

What the public endpoints actually do

We checked the site's current service code and its published OpenAPI description. The documented GET /api/agent route returns service guidance and can report whether a URL matches a Reel link pattern. That pattern check does not fetch the post or recognize speech. The POST /audio route retrieves audio for the browser tool; its response is audio data, not a transcript.

Speech recognition then runs inside the visitor's browser with the English Whisper model. The resulting words are placed in the page's result area, where Copy and Download .txt act on them. The current public API description has no link-in/text-out transcript operation. The site's MCP tools likewise provide guidance and URL-pattern validation, not a transcription result. This is a statement about XAPN's inspected implementation on that date, not every possible third-party service or future version.

The closest usable workflow now

  1. Copy a public Reel's link from Instagram. A profile, Story, or private post is not a valid substitute.
  2. Open XAPN's browser tool, paste the link, and select Get transcript. This is the manual step a shortcut cannot currently skip.
  3. Wait for Done. and visible words. If the page stops while fetching, decoding, or loading the model, do not treat a blank result as a transcript.
  4. Use Copy to move the result into notes. Review it against the audio before quoting it; recognition can mishear names or jargon.

If your requirement is one action in a phone shortcut that returns a text variable without opening a browser or pressing a button, this tool does not meet it today. Do not wire the URL-validation API or audio endpoint into a shortcut and label their responses a transcript.

Questions

Can a shortcut at least collect the Reel URL? A phone shortcut may help you copy or hand off a link, but we did not build or test an iOS or Android shortcut here. XAPN's verified flow starts when the link is pasted into its web page.

Does the tool read Instagram's native captions? The inspected flow downloads audio and recognizes speech in the browser. It is not a native-caption extraction endpoint. A Reel without spoken words cannot gain speech from an on-screen caption alone. A profile link also cannot produce every Reel transcript or written caption.

What if the browser run fails? Note the exact stage shown above the result and use the failure-stage guide. Repeating a URL-pattern check will not diagnose an audio-fetch or model problem.

A shortcut cannot bypass an unreachable or unsupported source. Use the Instagram to text page for the actual one-link conversion step. Automate only the handoff steps that remain predictable.