Is Atlassian deprecating v2 APIs entirely?

[Samrat Dutta]

While Atlassian has not fully phased out the Jira v2 APIs yet, some endpoints (such as Search and JQL) have already been deprecated.

I'd like to confirm whether this API is expected to remain supported in the near term. If deprecation is planned, it would be helpful to understand the timeline so we have sufficient runway to prepare and avoid any surprises for our customers.

[Artem Nek_Votazz · Accepted solution]

Hi Samrat

Short answer: no — Atlassian isn't retiring the v2 (or v3) Jira platform REST
API as a whole. Both remain supported; v3 is just the ADF-aware twin of v2,
not a replacement that deprecates it.

What got deprecated was a specific set of endpoints — not a version. This was
announced back in October 2024:

• GET/POST /rest/api/{2|3|latest}/search
• POST /rest/api/{2|3|latest}/search/id
• POST /rest/api/{2|3|latest}/expression/eval

Replacements:
• Search → POST /rest/api/{2|3|latest}/search/jql
• Expression eval → POST /rest/api/3/expression/evaluate

This one's already done, not upcoming: those endpoints went through their
sunset over 2025 and the old /search now returns 410 Gone on Cloud. So if
something is still pointed at them today, it's already broken rather than
"about to break" — migrating to /search/jql isn't optional anymore.

Two behavioural changes that catch people out on the new endpoint:
• queries are bounded — you must pass an explicit `fields` list (and
maxResults); no more implicit "return everything"
• pagination is cursor-based via `nextPageToken`, not startAt/total

For your underlying question about runway: there's no blanket v2/v3 retirement
to plan around. Deprecations are handled per-endpoint, each with its own
notice
period, and they all land in the Jira Cloud platform changelog — watching that
is the cleanest way to catch anything before it affects customers.

• How to run JQL search on the new endpoint (fields + nextPageToken
examples)
• Jira Cloud platform changelog:
https://developer.atlassian.com/cloud/jira/platform/changelog/

(Aside: if "v2" ever comes up in a Confluence context, that's the opposite
case
— there it's v1 being retired in favour of v2, so v2 is the one to be on.)

[Germán Morales _ Hiera · Accepted solution]

Hi @samrat, there are two separate things bundled into your question, and splitting them apart makes the answer a lot clearer.

On the broad point: no, Atlassian is not retiring the v2 (or v3) Jira platform REST API as a version. Both remain supported, and v3 is the ADF-aware counterpart of v2 rather than a replacement that obsoletes it. What gets deprecated is individual endpoints, each with its own removal date published in the developer changelog. There is no single shut-off date for the whole API that you need to plan around.

On the part you actually care about, the search endpoints: this is not a future plan, it has already happened on Cloud. The old GET/POST /rest/api/3/search (and the /2 and /latest equivalents) has been removed and now returns 410 Gone, so anything still pointed at it is already broken rather than about to break. The official note plus a working request example is here: Run JQL search query using Jira Cloud REST API.

The replacement is POST /rest/api/3/search/jql. Two changes tend to trip people up during the migration:

• You have to pass an explicit fields array. The new endpoint does not return every field by default.
• Paging is cursor-based via nextPageToken; startAt and the total count are gone. If all you need is a count, use the approximate-count operation rather than paging through everything (see the Issue search API group).

While you are auditing, the same notice also covered POST /rest/api/3/search/id and POST /rest/api/3/expression/eval, so it is worth grepping for those as well. The full announcement with the exact dates is changelog entry CHANGE-2046.

For the runway concern going forward: that developer changelog is the canonical source. Every deprecation lands there with a removal date before it takes effect, and it publishes an RSS feed you can filter and pipe into Slack, which is the cleanest way to catch anything that would reach your customers. One caveat worth stating: everything above is Jira Cloud specific. If any of your customers run Data Center, that platform follows its own lifecycle and these dates do not apply to them.

If you tell me whether your integration makes direct REST calls or goes through a client library, I can point you at the exact pieces that need changing.