Jira API: INVALID_INPUT error only in production when sending sanitized ADF payload

[Ad Ner]

Hi, I'm creating  issues through the REST API (POST /rest/api/3/issue) using an ADF
description. The request works correctly in my local and
staging environments, but in production it consistently returns:

400 Bad Request
{
"error_message": ["INVALID_INPUT"],
"errors": {}
}

I want to avoid posting sensitive data, so here are the details in a sanitized form:

What I have already checked:
- permissions: scope includes write access and logs confirm permission is OK
- Project permissions: the app/user can create issues in the target project
- The issue type exists and is the same as local/staging
- Same payload structure works in other environments

ADF sanitation before sending:
I sanitize the entire ADF tree to remove invalid nodes, empty nodes, and ensure correct
structure. Example of the sanitation method (simplified and sanitized):

sanitize (node):

• skip nodes without "type"

• recursively sanitize content children

• remove empty non-text nodes

• return cleaned node

I also log the maximum depth before sending, and depth never exceeds 4.

Additional notes:
- The same sanitized payload works in local and staging, but fails with INVALID_INPUT
only in production.
- The production payload is identical (no environment-specific differences).
- No custom ADF nodes or marks are being used.

My questions:
1. Under what conditions would Jira return INVALID_INPUT only in one environment if the
payload is identical?
2. Are there ADF constraints (depth limits, node types, empty content rules) that may
differ based on Jira Cloud tenant or configuration?
3. Is there a recommended validator or tool to identify which exact node is causing
INVALID_INPUT when the error response has no field details?
4. Could rate limits, app installation state, or partial app permissions cause this
specific error?

I can provide fully sanitized sample ADF snippets if helpful.

Thanks in advance!

[Sackmann_ Elias]

Hi @Ad Ner and thank you for your question.

I've encountered a similar issue before, where I received the same INVALID_INPUT error when creating Jira issues with ADF-formatted descriptions. In my case, the problem was caused by payload escaping (unescaped special characters or improperly structured JSON).

In my case, the ADF Renderer provided by Atlassian was pretty useful, you can find it here: Example - Basic - Renderer - Atlassian Frontend Development. Please note that it only renders your ADF and does not escape the payload for you.

With the proper escaping it worked. Unfortunately, since I’ve deleted the automation that had the issue, I can’t recover exact reproduction steps.

As far as I remember rate limitations and permission schemes were never an issue, however I can imagine that you need to have Editing permission within the space you want to publish in.

Hope this helps!