[ML] AI/Inference Connector creation: use 'location' field to correctly set provider config by alvarezmelissa87 · Pull Request #250838 · elastic/kibana

alvarezmelissa87 · 2026-01-28T23:26:58Z

Summary

Related issue: #241872

In upcoming changes, the services API will specify where a field should be included (service_settings or task_settings) - this will be an optional location field (similar to supported_task_types and how it's used). This will help Kibana build the request and place the fields in the correct settings object.

This PR ensures that field is used to set the config fields in the correct location - if the field is not present, then the default location should be service_settings.

This PR utilizes the existing but unused taskTypeConfig for keeping track of what fields end up going in task_settings. Anything in 'providerConfig' remains in service_settings.

The serializers have been updated to handle the old saved object format that didn't include taskTypeConfig.

Example of what the field will look like:

{
    "service": "openai",
    "name": "OpenAI",
    "task_types": [
        "text_embedding",
        "completion",
        "chat_completion"
    ],
    "configurations": {
        "api_key": {
            "description": "The OpenAI API authentication key. For more details about generating OpenAI API keys, refer to the https://platform.openai.com/account/api-keys.",
            "label": "API Key",
            "required": true,
            "sensitive": true,
            "updatable": true,
            "type": "str",
            "supported_task_types": [
                "text_embedding",
                "completion",
                "chat_completion"
            ]
        },
        "url": {
            "description": "The absolute URL of the external service to send requests to.",
            "label": "URL",
            "required": false,
            "sensitive": false,
            "updatable": false,
            "type": "str",
            "supported_task_types": [
                "text_embedding",
                "completion",
                "chat_completion"
            ],
         "location": a string either task_settings|service_settings 
        },
...
}

TESTING

To test, create a connector with Open AI as the provider (gpt-4.1 model id can be used) with custom headers on main (so the old saved object format is created) and create one on the branch to ensure both show up correctly. The same can be done for Anthropic (claude-sonnet-4-5 model id can be used) with max_tokens set.

For adaptive allocations, test on serverless and ensure that adaptive allocations are sent correctly for elastic/elasticsearch provider. On stateful it will be num_threads and num_allocations. You can create a connector/inference endpoint on main and then on the branch to ensure both show up correctly. Ping me if you need any api keys to use.

Checklist

Check the PR satisfies following conditions.

Reviewers should verify this PR satisfies this list as well.

Any text added follows EUI's writing guidelines, uses sentence case text and includes i18n support
Documentation was added for features that require explanation or tutorials
Unit or functional tests were updated or added to match the most common scenarios
If a plugin configuration key changed, check if it needs to be allowlisted in the cloud and added to the docker list
This was checked for breaking HTTP API changes, and any breaking changes have been approved by the breaking-change committee. The release_note:breaking label should be applied in these situations.
Flaky Test Runner was used on any tests changed
The PR description includes the appropriate Release Notes section, and the correct release_note:* label is applied per the guidelines
Review the backport guidelines and apply applicable backport:* labels.

github-actions · 2026-01-28T23:30:22Z

🤖 GitHub comments

Expand to view the GitHub comments

Just comment with:

/oblt-deploy : Deploy a Kibana instance using the Observability test environments.
run docs-build : Re-trigger the docs validation. (use unformatted text in the comment!)

pgayvallet

(on behalf of search-kibana, didn't full review)

alvarezmelissa87 · 2026-02-02T15:41:40Z

@elasticmachine merge upstream

qn895 · 2026-02-02T17:34:48Z

x-pack/platform/plugins/shared/inference_endpoint/server/routes/index.ts

          };

+          const taskSettingsWithHeaders = {
+            ...unflattenObject(config?.taskTypeConfig ?? {}), // TODO: confirm this unflatten is needed


Is this todo still needed?

alvarezmelissa87 · 2026-02-05T19:42:05Z

This PR is blocked by this bug fix (#251357) that needs to go in first.

alvarezmelissa87 · 2026-02-05T19:42:11Z

@elasticmachine merge upstream

elasticmachine · 2026-02-05T19:42:16Z

merge conflict between base and head

elasticmachine · 2026-02-10T13:12:57Z

Pinging @elastic/ml-ui (:ml)

x-pack/platform/packages/shared/kbn-inference-endpoint-ui-common/src/constants.tsx

alvarezmelissa87 · 2026-02-17T20:07:52Z

@elasticmachine merge upstream

alvarezmelissa87 · 2026-02-19T19:29:51Z

@elasticmachine merge upstream

peteharverson

LGTM.

Tested using Anthropic (claude-sonnet-4-5) and Elasticsearch service - created on main and on the PR branch, and can view / edit connectors successfully when running on your PR branch.

alvarezmelissa87 · 2026-03-11T17:58:56Z

@saikatsarkar056 - updated this with latest from main. Would you be up for giving this a test on review? 🙏

saikatsarkar056 · 2026-03-11T23:25:29Z

I did the following tests:

Test 1: OpenAI with Custom Headers

Go to Stack Management > Connectors > Create Connector > AI/Inference
Select OpenAI as the provider
Set the URL to https://openrouter.ai/api/v1/chat/completions
Set Model ID to openai/gpt-4.1
Enter your OpenRouter API key
Add a custom header (e.g. key: x-test, value: hello)
Click Save
In the Network tab, find the connector response
Verify:
- headers is inside taskTypeConfig (not providerConfig)
- model_id, url are inside providerConfig
Close and re-open the connector -- verify the header displays correctly

Test 2: Anthropic with max_tokens

Create another AI/Inference connector
Select Anthropic as the provider
Set the URL to https://openrouter.ai/api/v1/messages
Enter your OpenRouter API key
Set max_tokens to 1024
Click Save
In the Network tab, find the connector response
Verify:
- max_tokens is inside taskTypeConfig (not providerConfig)
- Other fields like model_id are inside providerConfig
Close and re-open the connector -- verify max_tokens displays correctly

saikatsarkar056 · 2026-03-12T15:08:04Z

...tform/plugins/shared/stack_connectors/public/connector_types/inference/form_serialization.ts

                  enabled: true,
                  min_number_of_allocations: MIN_ALLOCATIONS,
-                  ...(maxAllocations ? { max_number_of_allocations: maxAllocations } : {}),
+                  ...(max_number_of_allocations ? { max_number_of_allocations } : {}),


max_number_of_allocations is already checked at line: 27. We can just put max_number_of_allocations here instead of ternary operation

Suggested change

...(max_number_of_allocations ? { max_number_of_allocations } : {}),

max_number_of_allocations,

@saikatsarkar056 - good catch! Updated in e0e21e1

saikatsarkar056 · 2026-03-12T15:26:22Z

...packages/shared/kbn-inference-endpoint-ui-common/src/components/inference_flyout_wrapper.tsx

+    ...restProviderConfig
+  } = providerConfig || {};

+  if (restProviderConfig) {


I think after destructuring providerConfig || {}, restProviderConfig will be at least an empty object {} which is always truthy. Could you please verify whether this is true? I will try to test. In that case, we can never reach line 62: return data

const providerConfig = undefined; const { a, ...rest } = providerConfig || {}; if (rest) { console.log('always reaches here:', rest); } else { console.log('never reaches here'); } output: "always reaches here:", {}

@saikatsarkar056 - thank you for taking a look 🙏 You're totally right! Good catch - updated in e0e21e1

alvarezmelissa87 · 2026-03-17T21:17:55Z

This has been updated with latest main and all comments have been addressed. I tested all cases with latest changes.
Would appreciate a final look/test. cc @saikatsarkar056, @qn895, @Samiul-TheSoccerFan

qn895

LGTM 🎉

Samiul-TheSoccerFan

LGTM

elasticmachine · 2026-03-18T23:09:19Z

⏳ Build in-progress

Buildkite Build
Commit: 73c4314
Kibana Serverless Image: docker.elastic.co/kibana-ci/kibana-serverless:pr-250838-73c43142bb8c
Elasticsearch Serverless Deployment

Failed CI Steps

Test Failures

[job] [logs] Scout: [ observability / uptime ] plugin / local-stateful-classic - MonitorDetails - navigates to monitor details and displays ping data
[job] [logs] Scout: [ platform / dashboard-stateful-classic ] plugin / local-stateful-classic - Sync colors - should be possible to disable color sync

History

cc @alvarezmelissa87

…d_agent_navigation2 * commit '9289d6b5502db245e645e190b0246554396c6c20': (34 commits) [api-docs] 2026-03-19 Daily api_docs build (elastic#258471) [Shared UX][DateRangePicker] Missing parts (elastic#258229) [Dashboard] Keep pinned_panels separate in read response (elastic#258444) Move inheritance: true to top level in .coderabbit.yml (elastic#258461) [DOCS] 9.3.2 Kibana release notes (elastic#257332) adds routing accept metric attribute to the cps metric (elastic#258168) [ML] AI/Inference Connector creation: use 'location' field to correctly set provider config (elastic#250838) [Lens] Add e2e test for legend list layout (elastic#258160) [SigEvents] Convert feature duplication evaluators to createPrompt pattern (elastic#256534) Add actionable-obs author to .coderabbit.yml (elastic#257922) [DOCS] 9.2.7 Kibana release notes (elastic#257331) Grant Serverless editor/viewer access to ES v2 indices (elastic#258384) [SigEvents][Evals] Rename terminology for KI features and KI queries (elastic#258361) [EDR Workflows][Osquery] Add shared table toolbar components and redesign saved queries list (elastic#258394) [Automatic Import V2] Upload samples using an existing index (elastic#258074) Add GET /inference_features route to expose feature registry (elastic#258044) fix additional fields not included (elastic#257625) [Discover] [Metrics] Add tier 2 journeys for Metrics in Discover E2E (elastic#255036) [Lens as code] Support correct X-Axis types in ES|QL visualizations (elastic#258159) Update APM (main) (elastic#254880) ...

…ly set provider config (elastic#250838) ## Summary Related issue: elastic#241872 In upcoming changes, the services API will specify where a field should be included (`service_settings` or `task_settings`) - this will be an optional `location` field (similar to `supported_task_types` and how it's used). This will help Kibana build the request and place the fields in the correct settings object. This PR ensures that field is used to set the config fields in the correct location - if the field is not present, then the default location should be `service_settings`. This PR utilizes the existing but unused `taskTypeConfig` for keeping track of what fields end up going in task_settings. Anything in 'providerConfig' remains in service_settings. The serializers have been updated to handle the old saved object format that didn't include taskTypeConfig. Example of what the field will look like: ``` { "service": "openai", "name": "OpenAI", "task_types": [ "text_embedding", "completion", "chat_completion" ], "configurations": { "api_key": { "description": "The OpenAI API authentication key. For more details about generating OpenAI API keys, refer to the https://platform.openai.com/account/api-keys.", "label": "API Key", "required": true, "sensitive": true, "updatable": true, "type": "str", "supported_task_types": [ "text_embedding", "completion", "chat_completion" ] }, "url": { "description": "The absolute URL of the external service to send requests to.", "label": "URL", "required": false, "sensitive": false, "updatable": false, "type": "str", "supported_task_types": [ "text_embedding", "completion", "chat_completion" ], "location": a string either task_settings|service_settings }, ... } ``` ### TESTING To test, create a connector with Open AI as the provider (gpt-4.1 model id can be used) with custom headers on main (so the old saved object format is created) and create one on the branch to ensure both show up correctly. The same can be done for Anthropic (claude-sonnet-4-5 model id can be used) with max_tokens set. For adaptive allocations, test on serverless and ensure that adaptive allocations are sent correctly for elastic/elasticsearch provider. On stateful it will be num_threads and num_allocations. You can create a connector/inference endpoint on main and then on the branch to ensure both show up correctly. Ping me if you need any api keys to use. ### Checklist Check the PR satisfies following conditions. Reviewers should verify this PR satisfies this list as well. - [ ] Any text added follows [EUI's writing guidelines](https://elastic.github.io/eui/#/guidelines/writing), uses sentence case text and includes [i18n support](https://github.com/elastic/kibana/blob/main/src/platform/packages/shared/kbn-i18n/README.md) - [ ] [Documentation](https://www.elastic.co/guide/en/kibana/master/development-documentation.html) was added for features that require explanation or tutorials - [ ] [Unit or functional tests](https://www.elastic.co/guide/en/kibana/master/development-tests.html) were updated or added to match the most common scenarios - [ ] If a plugin configuration key changed, check if it needs to be allowlisted in the cloud and added to the [docker list](https://github.com/elastic/kibana/blob/main/src/dev/build/tasks/os_packages/docker_generator/resources/base/bin/kibana-docker) - [ ] This was checked for breaking HTTP API changes, and any breaking changes have been approved by the breaking-change committee. The `release_note:breaking` label should be applied in these situations. - [ ] [Flaky Test Runner](https://ci-stats.kibana.dev/trigger_flaky_test_runner/1) was used on any tests changed - [ ] The PR description includes the appropriate Release Notes section, and the correct `release_note:*` label is applied per the [guidelines](https://www.elastic.co/guide/en/kibana/master/contributing.html#kibana-release-notes-process) - [ ] Review the [backport guidelines](https://docs.google.com/document/d/1VyN5k91e5OVumlc0Gb9RPa3h1ewuPE705nRtioPiTvY/edit?usp=sharing) and apply applicable `backport:*` labels. --------- Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>

use taskTypeConfig for taskSetting fields

9a37e5d

alvarezmelissa87 mentioned this pull request Jan 28, 2026

[ML] AI/Inference Connector creation: use 'location' field to correctly set provider config #247307

Closed

8 tasks

alvarezmelissa87 self-assigned this Jan 28, 2026

alvarezmelissa87 added Feature: AI Infra Relating to the AI Assistant flow and any work impacting/involving the AI/Inference Connector v9.4.0 Feature:Inference UI ML Inference endpoints UI and AI connector release_note:enhancement labels Jan 28, 2026

alvarezmelissa87 marked this pull request as ready for review January 28, 2026 23:29

alvarezmelissa87 requested review from a team as code owners January 28, 2026 23:29

alvarezmelissa87 added ci:project-deploy-elasticsearch Create an Elasticsearch Serverless project ci:project-deploy-observability Create an Observability project labels Jan 28, 2026

alvarezmelissa87 added the backport:skip This PR does not require backporting label Jan 28, 2026

alvarezmelissa87 requested review from Samiul-TheSoccerFan, peteharverson, qn895 and saikatsarkar056 January 28, 2026 23:58

pgayvallet approved these changes Jan 29, 2026

View reviewed changes

alvarezmelissa87 added 2 commits January 29, 2026 11:39

fix serialization jest test

d8cce43

ensure validation works for secret key field

e9fcf21

Merge branch 'main' into inference-connector-location-handling

6d469fe

qn895 reviewed Feb 2, 2026

View reviewed changes

peteharverson added the :ml label Feb 10, 2026

peteharverson reviewed Feb 10, 2026

View reviewed changes

x-pack/platform/packages/shared/kbn-inference-endpoint-ui-common/src/constants.tsx Show resolved Hide resolved

elasticmachine and others added 2 commits February 17, 2026 21:07

Merge branch 'main' into inference-connector-location-handling

b631d8c

fix merge type and adaptive allocations deserializer

bfce342

Merge branch 'main' into inference-connector-location-handling

c625c5b

peteharverson approved these changes Feb 23, 2026

View reviewed changes

Merge branch 'main' into inference-connector-location-handling

ae256c5

saikatsarkar056 reviewed Mar 12, 2026

View reviewed changes

alvarezmelissa87 added 2 commits March 17, 2026 10:17

Merge branch 'main' into inference-connector-location-handling

0abcc93

clean up serializer

e0e21e1

saikatsarkar056 approved these changes Mar 17, 2026

View reviewed changes

qn895 approved these changes Mar 18, 2026

View reviewed changes

Samiul-TheSoccerFan approved these changes Mar 18, 2026

View reviewed changes

type fix for form serializer/deserializer

73c4314

neptunian approved these changes Mar 18, 2026

View reviewed changes

spong approved these changes Mar 18, 2026

View reviewed changes

alvarezmelissa87 merged commit c462948 into elastic:main Mar 18, 2026
18 checks passed

alvarezmelissa87 deleted the inference-connector-location-handling branch March 18, 2026 23:38

peteharverson mentioned this pull request Mar 19, 2026

[ML][AI Connector] Creation flyout: handle new 'location' field in configuration fields #241872

Closed

	...(max_number_of_allocations ? { max_number_of_allocations } : {}),
	max_number_of_allocations,

Conversation

alvarezmelissa87 commented Jan 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

TESTING

Checklist

Uh oh!

github-actions bot commented Jan 28, 2026

🤖 GitHub comments

Uh oh!

pgayvallet left a comment

Choose a reason for hiding this comment

Uh oh!

alvarezmelissa87 commented Feb 2, 2026

Uh oh!

qn895 Feb 2, 2026

Choose a reason for hiding this comment

Uh oh!

alvarezmelissa87 commented Feb 5, 2026

Uh oh!

alvarezmelissa87 commented Feb 5, 2026

Uh oh!

elasticmachine commented Feb 5, 2026

Uh oh!

elasticmachine commented Feb 10, 2026

Uh oh!

Uh oh!

alvarezmelissa87 commented Feb 17, 2026

Uh oh!

alvarezmelissa87 commented Feb 19, 2026

Uh oh!

peteharverson left a comment

Choose a reason for hiding this comment

Uh oh!

alvarezmelissa87 commented Mar 11, 2026

Uh oh!

saikatsarkar056 commented Mar 11, 2026

Test 1: OpenAI with Custom Headers

Test 2: Anthropic with max_tokens

Uh oh!

saikatsarkar056 Mar 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

alvarezmelissa87 Mar 17, 2026

Choose a reason for hiding this comment

Uh oh!

saikatsarkar056 Mar 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

alvarezmelissa87 Mar 17, 2026

Choose a reason for hiding this comment

Uh oh!

alvarezmelissa87 commented Mar 17, 2026

Uh oh!

qn895 left a comment

Choose a reason for hiding this comment

Uh oh!

Samiul-TheSoccerFan left a comment

Choose a reason for hiding this comment

Uh oh!

elasticmachine commented Mar 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

⏳ Build in-progress

Failed CI Steps

Test Failures

History

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

9 participants

alvarezmelissa87 commented Jan 28, 2026 •

edited

Loading

saikatsarkar056 Mar 12, 2026 •

edited

Loading

saikatsarkar056 Mar 12, 2026 •

edited

Loading

elasticmachine commented Mar 18, 2026 •

edited

Loading