[Obs AI Assistant] Migrate anonymization to inference by neptunian · Pull Request #225539 · elastic/kibana

neptunian · 2025-06-26T19:00:43Z

Migrates anonymization functionality from observability ai assistant to inference plugin
Migrates the anonymization rules advanced / ui setting to inference plugin
Changes advanced setting to aiAssistant:anonymizationRules and adds allowedEntityClasses for users to specify which entity types to anonymize from the NER model

code contribution models: claud 3.7 sonnet, o3

...k/platform/plugins/shared/inference/server/chat_complete/anonymization/anonymize_messages.ts

.../platform/plugins/shared/inference/server/chat_complete/anonymization/deanonymize_message.ts

dgieselaar · 2025-06-27T15:50:59Z

...ared/observability_ai_assistant/server/service/client/operators/get_deanonymized_messages.ts

+
+        return result;
+      }),
+      defaultIfEmpty(messages)


why do we need this?

the filter is strict, checking for deanonymized_input or deanonymized_output and if nothing passes it, nothing will emit without this. I could broaden the filter and add some conditions and return the message in there if that's a preferred style / safer. I thought this was a bit more readable.

x-pack/platform/plugins/shared/observability_ai_assistant/common/types.ts

dgieselaar · 2025-06-27T16:07:56Z

x-pack/platform/plugins/shared/observability_ai_assistant/common/types.ts

+    toolCalls?: Array<{
+      toolCallId: string;
+      function: {
+        name: string;
+        arguments: Record<string, any>;
+      };
+    }>;


I get confused about the data modeling here. I think this should either just be the output from the inference client, or it should be our own model (which means single function_call).

Isn't that the output from the inference client? array of toolCalls.

what I mean is you have mixed up the types from the inference client (indeed, multiple tool calls), and our format: name is ours, toolCallId is from the inference client, but they serve the same purpose (identifying the tool call request). We can leave it for later but we should fix it.

elasticmachine · 2025-06-27T18:01:13Z

Pinging @elastic/obs-ai-assistant (Team:Obs AI Assistant)

github-actions · 2025-06-27T18:01:18Z

🤖 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!)

…om/neptunian/kibana into migrate-anonymization-to-inference

dgieselaar · 2025-07-04T17:19:43Z

@elasticmachine merge upstream

…zation-to-inference

elasticmachine · 2025-07-07T09:00:36Z

💚 Build Succeeded

Buildkite Build
Commit: fa1161d
Kibana Serverless Image: docker.elastic.co/kibana-ci/kibana-serverless:pr-225539-fa1161d44fe7

Metrics [docs]

Module Count

Fewer modules leads to a faster build time

id	before	after	diff
`automaticImport`	796	797	+1
`elasticAssistant`	451	452	+1
`inference`	48	49	+1
`observabilityAIAssistant`	108	107	-1
`observabilityAIAssistantApp`	434	435	+1
`observabilityAiAssistantManagement`	265	266	+1
`productDocBase`	40	41	+1
`searchAssistant`	151	152	+1
`searchPlayground`	366	367	+1
`securitySolution`	7819	7820	+1
`workchatApp`	181	182	+1
total			+9

Public APIs missing comments

Total count of every public API that lacks a comment. Target amount is 0. Run node scripts/build_api_docs --plugin [yourplugin] --stats comments for more detailed information.

id	before	after	diff
`@kbn/inference-common`	130	161	+31
`observabilityAIAssistant`	438	429	-9
total			+22

Async chunks

Total size of all lazy-loaded chunks that will be downloaded as the user navigates the app

id	before	after	diff
`observabilityAIAssistantApp`	176.1KB	176.1KB	-56.0B
`searchAssistant`	102.9KB	102.8KB	-56.0B
total			-112.0B

Public APIs missing exports

Total count of every type that is part of your API that should be exported but is not. This will cause broken links in the API documentation system. Target amount is 0. Run node scripts/build_api_docs --plugin [yourplugin] --stats exports for more detailed information.

id	before	after	diff
`observabilityAIAssistant`	32	30	-2

Page load bundle

Size of the bundles that are downloaded on every page load. Target size is below 100kb

id	before	after	diff
`observabilityAIAssistant`	41.9KB	41.4KB	-553.0B

Unknown metric groups

API count

id	before	after	diff
`@kbn/inference-common`	271	302	+31
`observabilityAIAssistant`	444	431	-13
total			+18

History

kibanamachine · 2025-07-07T10:36:51Z

Starting backport for target branches: 8.19, 9.1

https://github.com/elastic/kibana/actions/runs/16114647517

- Migrates anonymization functionality from observability ai assistant to inference plugin - Migrates the anonymization rules advanced / ui setting to inference plugin - Changes advanced setting to `aiAssistant:anonymizationRules` and adds `allowedEntityClasses` for users to specify which entity types to anonymize from the NER model code contribution models: claud 3.7 sonnet, o3 --------- Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com> Co-authored-by: Dario Gieselaar <dario.gieselaar@elastic.co> Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com> (cherry picked from commit b1bdc6e)

kibanamachine · 2025-07-07T10:43:58Z

💔 Some backports could not be created

Status	Branch	Result
❌	8.19	Backport failed because of merge conflicts You might need to backport the following PRs to 8.19: - [Obs AI Assistant] Collapse *query tool calls (#226078) - Move test utils from legacy api to agnostic (#224381)
✅	9.1

Note: Successful backport PRs will be merged automatically after passing CI.

Manual backport

To create the backport manually run:

node scripts/backport --pr 225539

Questions ?

Please refer to the Backport tool documentation

…#226763) # Backport This will backport the following commits from `main` to `9.1`: - [[Obs AI Assistant] Migrate anonymization to inference (#225539)](#225539)  ### Questions ? Please refer to the [Backport tool documentation](https://github.com/sorenlouv/backport)  Co-authored-by: Sandra G <neptunian@users.noreply.github.com> Co-authored-by: Dario Gieselaar <dario.gieselaar@elastic.co> Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>

- Migrates anonymization functionality from observability ai assistant to inference plugin - Migrates the anonymization rules advanced / ui setting to inference plugin - Changes advanced setting to `aiAssistant:anonymizationRules` and adds `allowedEntityClasses` for users to specify which entity types to anonymize from the NER model code contribution models: claud 3.7 sonnet, o3 --------- Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com> Co-authored-by: Dario Gieselaar <dario.gieselaar@elastic.co> Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com> (cherry picked from commit b1bdc6e) # Conflicts: # x-pack/platform/plugins/shared/observability_ai_assistant/server/service/client/index.ts

neptunian · 2025-07-10T20:52:26Z

💚 All backports created successfully

Status	Branch	Result
✅	8.19

Note: Successful backport PRs will be merged automatically after passing CI.

Questions ?

Please refer to the Backport tool documentation

#227543) # Backport This will backport the following commits from `main` to `8.19`: - [[Obs AI Assistant] Migrate anonymization to inference (#225539)](#225539)  ### Questions ? Please refer to the [Backport tool documentation](https://github.com/sorenlouv/backport)

## Summary Followup to #225539 Adds an anonymization-regex worker pool to run user-defined regex patterns out-of-process, protecting the Node event-loop from catastrophic back-tracking. For internal use we have config to adjust or turn off the worker: • `enabled` – turn worker on/off • `minThreads` / `maxThreads` – worker size • `idleTimeout` – how long a worker thread is allowed to be idle • `taskTimeout` – per-regex task limit Contributions by windsurf(o3).

## Summary Followup to elastic#225539 Adds an anonymization-regex worker pool to run user-defined regex patterns out-of-process, protecting the Node event-loop from catastrophic back-tracking. For internal use we have config to adjust or turn off the worker: • `enabled` – turn worker on/off • `minThreads` / `maxThreads` – worker size • `idleTimeout` – how long a worker thread is allowed to be idle • `taskTimeout` – per-regex task limit Contributions by windsurf(o3). (cherry picked from commit 1b9063a)

- Migrates anonymization functionality from observability ai assistant to inference plugin - Migrates the anonymization rules advanced / ui setting to inference plugin - Changes advanced setting to `aiAssistant:anonymizationRules` and adds `allowedEntityClasses` for users to specify which entity types to anonymize from the NER model code contribution models: claud 3.7 sonnet, o3 --------- Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com> Co-authored-by: Dario Gieselaar <dario.gieselaar@elastic.co> Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>

## Summary Followup to elastic#225539 Adds an anonymization-regex worker pool to run user-defined regex patterns out-of-process, protecting the Node event-loop from catastrophic back-tracking. For internal use we have config to adjust or turn off the worker: • `enabled` – turn worker on/off • `minThreads` / `maxThreads` – worker size • `idleTimeout` – how long a worker thread is allowed to be idle • `taskTimeout` – per-regex task limit Contributions by windsurf(o3).

neptunian added 5 commits June 26, 2025 13:48

migrate assistant to inference

f4b8667

remove anonymization instructions for now

6fc381d

fix merge conflict

c35c61d

unredactions => anonymizations

aac0d56

fix tests

863f750

neptunian added release_note:skip Skip the PR/issue when compiling release notes backport:version Backport to applied version labels labels Jun 27, 2025