Instagram API for agents: public profile and post data

Someone gives you a list of Instagram accounts and asks which ones post often, which posts got the most comments, and what the captions say. You open the Instagram API docs and find that most of it is built for managing your own account, not for reading other people's public posts.

That gap is why people search for an Instagram API or an Instagram scraper. This post shows how an agent pulls public Instagram data through the looot catalog, what comes back, what a call costs, and where the data stops. It also says plainly what you should not collect.

What does an Instagram API call return?

A call to the user posts endpoint returns a list of a public account's recent posts as JSON. A post typically carries the caption, the media type, the like and comment counts, the post URL and the time it was published, though the exact fields depend on the provider.

You send a username or, for some providers, a numeric user id. The endpoint answers with one page of posts. Several providers offer the same job, so the catalog lists them side by side with a price and, where one exists, a measured success rate.

Other Instagram endpoints may exist, such as profile details or comments. Search the catalog for them and read what each one lists. Do not assume a field exists. Inspect the endpoint, which is free, and read its output before you build on it.

How do I get Instagram data without writing a scraper?

You ask your agent to search the catalog for the job and run the endpoint it finds. The provider does the collecting. You send a username and read a table back, with no login to keep alive and no page parsing.

The agent follows four steps:

1. Search the catalog for "instagram user posts". Searching is free.
2. Inspect the cheapest endpoint that fits. Read its inputs, its output fields and its price.
3. Run it on one account and read the raw result.
4. Run the list, saving one row per post with the endpoint and cost.

The one-line setup connects the agent. A new workspace starts at zero balance and there is no trial credit, so top up before the first run.

What does it cost to pull Instagram posts?

The price is per call, and one call returns one page of one account's posts. So the cost of a job is the number of accounts times the pages you want. The table reads the cheapest priced endpoint for the job from the live catalog when the page loads.

To turn a per call price into a batch price, use the method in what 1,000 lookups cost. The short version is to multiply, then add a margin for the pages that return nothing.

What can an agent do with Instagram data?

The jobs that pay off are the ones that compare many accounts at once.

• Competitor content review. Pull the last 30 posts of five accounts and ask the agent which formats and topics got the most comments. It reads captions faster than you do.
• Creator shortlist. For a list of candidate creators, pull recent posts and compute the average comment count per post. The recipe for an influencer shortlist lays out the steps.
• Caption language. Collect the captions of top posts in a niche and have the agent list the hooks that repeat. Use the list as input, not as copy to paste.

Like counts and view counts are what the provider returns at call time, and platforms sometimes hide them. Treat a missing count as unknown, not as zero.

Instagram scraper or Instagram API: what is the difference here?

You get JSON, so it works like an API. The provider collects public pages and fields on its side, so it is a scraper under the hood. You never hold an Instagram login and you never run browser automation.

That has a cost. A provider can lose access to a field, and your agent may see a call fail after it worked yesterday. Build the job so one failed call does not stop the batch: ask the agent to record the failure and move on. And keep to public data. Do not collect private accounts, direct messages or anything behind a login, and read the platform's terms before you build a product on the data.

How do I check an Instagram result before I trust it?

Run the same account through two providers and compare. Pick one account you know well, call the cheapest endpoint, then call the second cheapest. If the post counts and the comment numbers agree, the data is usable. If they differ, open the account in a browser and see which provider is right.

Do this once per endpoint, not once per job. It costs a few cents and it tells you whether the cheapest provider is also a good one. The catalog shows a measured success rate for some endpoints, which tells you how often a call worked, not whether the numbers were right. Only your spot check answers that.

Write down which provider you chose and why, in the prompt or in a note. Three months from now someone will ask why the numbers changed, and the answer is usually that the provider did.

Run it: a prompt for your agent

Paste this into an agent that can run shell commands.

What it cannot do

• It reads public data only. It does not post, like, follow or message.
• It does not read private accounts or stories that need a login.
• It does not run on a timer. Checking an account on a schedule needs scheduled runs, not available yet, so rerun it by hand.
• It does not return your own account's analytics. That is a different job and a different kind of access.
• Fields vary by provider. Some return view counts, some do not. Inspect before you rely on one.

The same pattern works for video data, as shown in the TikTok guide. To browse every Instagram endpoint and its price, open the public catalog.

Last checked 2026-10-03 against the live catalog. Author: Walid Boulanouar.