Instagram Stories Scraper

Instagram Stories Scraper

Extract the currently available stories from public Instagram accounts. Runs work with no setup: no login, no cookie, no app review, nothing to supply.

It returns story metadata and direct media URLs. It does not download media binaries, read private accounts, or reach expired stories or viewer lists.

What you need to know before you run it

Story trays work out of the box. A story is ephemeral and never indexed, so Instagram serves a live tray only to a signed-in caller — there is no logged-out route to it at any price. This actor reads the tray through a licensed data source. You supply nothing.

A run is a snapshot. Stories expire about 24 hours after they are posted, so you get exactly what was live at the moment the run happened, and nothing older. There is no archive to go back to. An account with no live stories right now returns nothing, and that is a correct answer, not an error — it arrives as an uncharged NO_CURRENT_STORIES row and costs you nothing.

Two fields the licensed source does not carry. This is the one real limitation and it is worth knowing before you start: on rows served by the licensed source, expiresAt and dimensions come back null. The source publishes no expiry timestamp — and a story's 24-hour lifetime is a rule of thumb, not data, so it is left empty rather than computed and sold to you as a fact — and the payload carries no media width or height at all. Every other field on the row is populated as documented. If you need those two fields, supply your own cookie: read directly from Instagram, they are present.

To use your own account — for those two fields, or to run on your own account's rate limit — put your cookie in sessionCookies:

• In Chrome: F12 → Application → Cookies → https://www.instagram.com → copy the sessionid

value. Pasting the whole cookie string works too.

• Your cookie takes priority over everything else. The run reads Instagram directly with it, and

reaches for the licensed source only for accounts that lane could not answer at all.

• A session cookie is equivalent to an active login. Use an account you own and are authorized to

automate, keep the value in Apify secrets rather than in a saved task, and revoke it through your normal account process when you are done.

Cookie values, proxy URLs, and request headers are never logged, written to the dataset, stored in this repository, or included in diagnostic rows.

Input

With empty input the actor returns exactly one clearly labeled _sample row. The sample makes no Instagram request, no proxy connection, no credential use, and no billing event.

Method and reliability

Without a cookie, each requested username is sent straight to the licensed source, which is addressed by username and returns the whole live tray in a single lookup. That means one lookup per account, whatever maxStoriesPerProfile is set to — the tray is not paged.

With your cookie, the username is first resolved to its numeric profile id from the public, logged-out profile endpoint, because a profile page answers a guest request and costs nothing; your cookie is spent on that step only if the guest attempt comes back gated. The tray itself is then read with the cookie, paired one-to-one with a sticky datacenter IP. A 403, 429, server error, or network failure rotates to a fresh sticky proxy session up to the configured bound. If Instagram challenges datacenter addresses, supply your own residential endpoints in proxyUrls — they are honoured exactly as given.

If that direct lane returns no stories at all, the run retries only the accounts it never got an answer for, through the licensed source. Accounts it did answer — including the ones that answered "nothing live right now" — are not looked up a second time, and no story can be billed twice: rows are deduplicated on storyId and charged once each.

Private accounts are out of reach on both paths, because nobody outside the follower list can read those stories. They are reported as an uncharged PRIVATE_OR_RESTRICTED row and no further lookup is spent on them.

Failures are named for what they are rather than flattened into one generic error: a rate limit reads as a rate limit, an upstream outage as an upstream outage, and a problem on our side is never reported as a problem on yours.

When a run stops early

A run can stop before every account is checked — the run deadline passes, or the actor reaches the per-run lookup budget set on the licensed source. That is a soft landing, never a crash: every story already collected is delivered and correct, and one uncharged PARTIAL or DEADLINE_EXCEEDED row tells you it stopped and why. Re-run to pick up the rest.

Story rows are written to the dataset before the billing event is charged. Samples, private-profile records, empty story trays, early stops, and all other diagnostics are never charged.

Output

Each genuine row contains:

• storyId, sourceUsername, sourceProfileId, and ownerUsername
• mediaType, imageUrl, videoUrl, and durationSeconds
• takenAt, caption, paid-partnership flag, and location when present
• expiresAt and dimensions — populated when the run used your own cookie, null on the licensed

source (see the limitation above)

• A canonical storyUrl

Stories are ephemeral. NO_CURRENT_STORIES means the lookup completed normally and the public account simply had nothing live at that moment.

Pricing

$0.40 per 1,000 stories ($0.0004 each), plus $0.001 each time a run starts. Flat rate — no volume tiers, no plan gates — and you are charged only for stories actually returned.

The billed event fires once per unique story row, and only after that row has been written to the dataset. Samples, private-profile records, empty story trays, early stops, and every other diagnostic are never charged — a run that returns no stories costs only the $0.001 start fee.