diff --git a/.github/skills/find-package-skill/SKILL.md b/.github/skills/find-package-skill/SKILL.md
new file mode 100644
index 000000000000..2dfb300c0357
--- /dev/null
+++ b/.github/skills/find-package-skill/SKILL.md
@@ -0,0 +1,25 @@
+---
+name: find-package-skill
+description: '**UTILITY SKILL** — Must be consulted before making any changes to packages under sdk/. Provides package-specific domain knowledge that prevents common mistakes. WHEN: "add feature to package", "fix bug in package", "modify package code", "regenerate from typespec", "change client library".'
+---
+
+# Find Package Skill
+
+Before modifying any SDK package in this monorepo, check this registry to see
+if a package-specific skill exists. Package skills contain tribal knowledge
+(architecture, data flows, type mappings, pitfalls) that prevents common mistakes.
+
+Always check this registry before modifying any SDK package — even if you think
+you already know the package well.
+
+## How to Use
+
+1. Find the package you're modifying in the table below.
+2. Read the SKILL.md at the listed path using the Read tool. Then read all files under the `references/` directory next to it for additional context.
+3. If the package isn't listed, no package-specific skill exists yet — proceed normally.
+
+## Package Skills
+
+| Package                   | Path                                                                        |
+| ------------------------- | --------------------------------------------------------------------------- |
+| `azure-search-documents` | `sdk/search/azure-search-documents/.github/skills/search-documents/SKILL.md` |
\ No newline at end of file
diff --git a/sdk/search/azure-search-documents/.github/skills/search-documents/SKILL.md b/sdk/search/azure-search-documents/.github/skills/search-documents/SKILL.md
new file mode 100644
index 000000000000..3b438d5940b1
--- /dev/null
+++ b/sdk/search/azure-search-documents/.github/skills/search-documents/SKILL.md
@@ -0,0 +1,224 @@
+---
+name: search-documents-python
+description: 'Post-regeneration customization guide for the azure-search-documents Python SDK. After running tsp-client update, consult this skill to re-apply all search-specific customizations and produce a production-ready SDK. USE WHEN: regenerating the search SDK from TypeSpec, running tsp-client update, fixing broken _patch.py imports after regeneration, adding a new operation or model to the search SDK, verifying the SDK still works after spec changes, or any task that mentions "regenerate", "tsp-client", "typespec", or "codegen" in the context of this package.'
+---
+
+# azure-search-documents — Post-Regeneration Customization Guide
+
+After running `tsp-client update`, the generated code is a raw skeleton — not a shippable SDK. This skill tells you exactly what customizations exist in `_patch.py` files and how to verify they still work after regeneration.
+
+The generator never touches `_patch.py` files, so your customizations survive regeneration. But they import from generated modules that **do** change. A renamed model, a new parameter, or a removed enum value will silently break a `_patch.py` file. The verification steps below catch these breaks.
+
+## Step 1: Run Regeneration
+
+```bash
+cd sdk/search/azure-search-documents
+
+# Update tsp-location.yaml with the new spec commit SHA, then:
+tsp-client update
+
+# If API version changed, update _metadata.json too
+```
+
+## Step 2: Verify Imports in All `_patch.py` Files
+
+Every `_patch.py` with customizations imports from generated modules. After regeneration, check that these imports still resolve. The fastest way:
+
+```bash
+python -c "from azure.search.documents import SearchClient"
+python -c "from azure.search.documents.aio import SearchClient"
+python -c "from azure.search.documents.indexes.models import SearchField, SearchFieldDataType"
+python -c "from azure.search.documents.models import IndexDocumentsBatch"
+```
+
+If any fail, a generated class/enum was renamed or removed. Read `references/customizations.md` for the exact import each `_patch.py` depends on.
+
+The `_patch.py` files that have customizations (all others are empty boilerplate):
+
+```
+azure/search/documents/
+├── _patch.py                    # SearchClient, SearchIndexingBufferedSender, ApiVersion
+├── _operations/_patch.py        # SearchItemPaged, search(), document CRUD
+├── models/_patch.py             # IndexDocumentsBatch, RequestEntityTooLargeError
+├── aio/_patch.py                # Async SearchClient, async SearchIndexingBufferedSender
+├── aio/_operations/_patch.py    # AsyncSearchItemPaged, async search()/CRUD
+├── indexes/_operations/_patch.py        # Polymorphic delete/update, list helpers
+├── indexes/models/_patch.py             # SearchField, field builders, enum aliases
+└── indexes/aio/_operations/_patch.py    # Async index/indexer operations
+```
+
+## Step 3: Check ApiVersion
+
+After regeneration, the generated code targets the API version in `_metadata.json`. The hand-maintained `ApiVersion` enum in `_patch.py` must include this version.
+
+```bash
+# 1. See what API version the generator used
+python -c "import json; print(json.load(open('_metadata.json'))['apiVersion'])"
+
+# 2. See what versions the SDK currently advertises
+grep -A 20 'class ApiVersion' azure/search/documents/_patch.py
+
+# 3. Check the current default
+grep 'DEFAULT_VERSION' azure/search/documents/_patch.py
+```
+
+If the generated API version is **not** in the `ApiVersion` enum:
+1. Add the new member to `ApiVersion` in `azure/search/documents/_patch.py` (e.g., `V2026_05_01_PREVIEW = "2026-05-01-preview"`)
+2. Update `DEFAULT_VERSION` to point to the new member
+3. Update the `ApiVersion` docstring in the class if one exists
+
+Verify the default round-trips correctly:
+
+```bash
+python -c "from azure.search.documents._patch import ApiVersion, DEFAULT_VERSION; print(DEFAULT_VERSION.value)"
+```
+
+## Step 4: Check for New Operations That Need Wrappers
+
+Look at what changed in the generated operations files:
+
+```bash
+git diff --name-only | grep "_operations\.py" | grep -v _patch
+```
+
+If a new operation was added to the generated `_operations.py`, decide whether it needs a convenience wrapper in `_patch.py`. Operations that need wrappers:
+- **Delete operations** — need the polymorphic str-or-model pattern (see "Polymorphic Delete Pattern" below)
+- **Create-or-update operations** — need `prefer="return=representation"`, `match_condition`, and `etag` forwarding
+- **List operations** — may need a `select` parameter or name-only projection
+
+Operations that pass through without a wrapper (if the generated signature is already user-friendly) can be left alone — they're inherited from the generated mixin.
+
+Remember: every sync wrapper in `_operations/_patch.py` needs a matching async wrapper in `aio/_operations/_patch.py`.
+
+## Step 5: Check for Model/Enum Changes That Affect Customizations
+
+```bash
+git diff models/_enums.py models/_models.py indexes/models/_enums.py indexes/models/_models.py
+```
+
+Watch for:
+- **Renamed enum values** — backward-compat aliases in `_patch.py` may reference old names
+- **New enum values that collide with Python keywords** — the generator renames them with `_ENUM` suffix (e.g., `IS` → `IS_ENUM`), and you must add a backward-compat alias
+- **Changed model constructors** — `IndexDocumentsBatch`, `SearchField`, `SearchIndexerDataSourceConnection`, `KnowledgeBase` are subclassed in `_patch.py` and may break if the base constructor changes
+- **New fields on `SearchResult`** — `_convert_search_result()` in `_operations/_patch.py` extracts `@search.*` metadata fields; new ones need to be added
+- **Changed `SearchRequest` model** — `_build_search_request()` constructs this model directly; new parameters need to be wired through
+- **Changed `SearchFieldDataType` model** — `indexes/models/_patch.py` monkey-patches `SearchFieldDataType.Collection` as a `staticmethod` and adds camelCase backward-compat aliases (e.g., `Int32` → `INT32`). If values are added, removed, or renamed in the generated enum, the aliases and `Collection` helper must be updated to match
+
+## Step 6: Ensure mypy pass
+
+```bash
+cd sdk/search/azure-search-documents
+
+# Preferred — uses azpysdk CLI (install via: pip install -e eng/tools/azure-sdk-tools)
+azpysdk mypy
+```
+
+The package-level `mypy.ini` already ignores `_generated` internals — you only need to fix errors in `_patch.py` files and `samples/`.
+
+## Step 7: Ensure pylint pass
+
+```bash
+cd sdk/search/azure-search-documents
+
+# Preferred
+azpysdk pylint
+```
+
+The repo-level `pylintrc` already excludes `_generated/`, `_vendor/`, `tests/`, and `samples/` — only `_patch.py` customizations are linted.
+
+## Step 8: Update Documentation and Samples
+
+If the regeneration added new operations, models, or changed the API version, update the docs and samples:
+
+### Changelog
+
+```bash
+# Edit CHANGELOG.md — add an entry under the next unreleased version
+# Format:
+#   ## 11.x.0bN (Unreleased)
+#   ### Features Added
+#   - Added `new_operation` to `SearchIndexClient`
+#   ### Breaking Changes
+#   - Renamed `OldModel` to `NewModel`
+code CHANGELOG.md
+```
+
+### README
+
+If new client classes or major features were added, update `README.md` with usage examples:
+
+```bash
+code README.md
+```
+
+# Customization Patterns Reference
+
+Each pattern below describes a customization that exists in the codebase today. After regeneration, verify each one still works. Read `references/customizations.md` for the exhaustive file-by-file inventory.
+
+## SearchClient Constructor Reordering
+
+`SearchClient` in `_patch.py` swaps parameter order to `(endpoint, index_name, credential)`. The generated base uses `(endpoint, credential, index_name)`. If regeneration changes the base constructor signature, update the subclass.
+
+## Custom Search Pagination
+
+`search()` does **not** use standard Azure SDK paging. It uses `SearchItemPaged` / `AsyncSearchItemPaged` with a custom `SearchPageIterator` because:
+- Search paginates via POST with `nextPageParameters` body, not GET with `nextLink`
+- First-page metadata (facets, count, answers) must be available before iteration
+- Results are converted to dicts with `@search.*` keys
+
+The continuation token is Base64-encoded JSON: `{"apiVersion": "...", "nextLink": "...", "nextPageParameters": {...}}`
+
+Shared helpers (`_build_search_request`, `_convert_search_result`, `_pack_continuation_token`, `_unpack_continuation_token`) live in sync `_operations/_patch.py` and are imported by async. Don't duplicate them.
+
+## Pipe-Delimited Semantic Encoding
+
+`_build_search_request()` encodes semantic parameters into pipe-delimited wire format:
+```
+query_answer="extractive", count=3  →  "extractive|count-3"
+query_caption="extractive", highlight=True  →  "extractive|highlight-true"
+```
+New semantic parameters should follow this pattern.
+
+## SearchIndexingBufferedSender
+
+Entirely hand-authored in `_patch.py` (sync) and `aio/_patch.py` (async). Not generated. Wraps `SearchClient` with auto-flush timer, 413 recursive batch splitting, retry per document key (409/422/503), and key field auto-detection. The async version supports both sync and async callbacks via `asyncio.iscoroutinefunction()`.
+
+## Field Builders
+
+`SimpleField()`, `SearchableField()`, `ComplexField()` in `indexes/models/_patch.py`. `SimpleField` explicitly sets `searchable=False`. `SearchableField` auto-sets type to `String`/`Collection(String)`.
+
+`SearchField` subclass adds `hidden` property (inverse of `retrievable`).
+
+`SearchFieldDataType.Collection` is a `staticmethod` monkey-patched onto the generated enum.
+
+## Backward-Compatible Enum Aliases
+
+Monkey-patched at module load in `_patch.py` files:
+```python
+SearchFieldDataType.Int32 = SearchFieldDataType.INT32   # camelCase → UPPER
+OcrSkillLanguage.IS = OcrSkillLanguage.IS_ENUM           # Python keyword collision
+ScoringStatistics.Global = ScoringStatistics.GLOBAL_ENUM
+```
+After regeneration, verify the right-hand-side names still exist in the generated enums. If a new enum collides with a Python keyword, add an alias.
+
+## Polymorphic Delete Pattern
+
+All delete/update operations in `indexes/_operations/_patch.py` accept str or model object:
+```python
+def delete_index(self, index, *, match_condition=MatchConditions.Unconditionally, **kwargs):
+    try:
+        name = index.name   # model object
+        return self._delete_index(name=name, etag=index.e_tag, match_condition=match_condition, **kwargs)
+    except AttributeError:
+        name = index         # string
+        return self._delete_index(name=name, **kwargs)
+```
+New resource types need this same wrapper in both sync and async.
+
+## Removed Model Tombstones
+
+`EntityRecognitionSkill`, `SentimentSkill` → `_RemovedModel` subclasses that `raise ValueError` on instantiation. Keep these for backward compat.
+
+## `_convert_index_response` Helper
+
+`list_indexes(select=...)` returns `SearchIndexResponse` (projection type). `_convert_index_response()` maps it to `SearchIndex`, notably `response.semantic` → `semantic_search`. Shared between sync and async via import.
diff --git a/sdk/search/azure-search-documents/.github/skills/search-documents/references/customizations.md b/sdk/search/azure-search-documents/.github/skills/search-documents/references/customizations.md
new file mode 100644
index 000000000000..ca2712246dd4
--- /dev/null
+++ b/sdk/search/azure-search-documents/.github/skills/search-documents/references/customizations.md
@@ -0,0 +1,232 @@
+# Post-Regeneration Customization Checklist
+
+Use this file after running `tsp-client update` to verify every customization. Each section is a `_patch.py` file with the exact classes, functions, and aliases it defines, plus what generated symbols it depends on.
+
+---
+
+## File: `azure/search/documents/_patch.py`
+
+### Depends On (from generated code)
+- `._client.SearchClient` as `_SearchClient` (base class)
+
+### Defines
+| Symbol | Type | What It Does |
+|--------|------|-------------|
+| `SearchClient` | class | Subclass of `_SearchClient`; reorders constructor to `(endpoint, index_name, credential)` |
+| `SearchIndexingBufferedSender` | class | Hand-authored batching sender. Uses `threading.Timer` for auto-flush, recursive 413 splitting, retry per doc key (409/422/503), key field auto-detection |
+| `ApiVersion` | enum | Supported API versions. Default: `V2025_11_01_PREVIEW` |
+| `is_retryable_status_code()` | function | Returns True for 409, 422, 503 |
+
+### After Regeneration, Verify
+- [ ] `_SearchClient` base class constructor signature unchanged
+- [ ] If API version changed, add new `ApiVersion` member and update default
+
+---
+
+## File: `azure/search/documents/models/_patch.py`
+
+### Depends On (from generated code)
+- `.._generated.models.IndexDocumentsBatch` as `IndexDocumentsBatchGenerated` (base class)
+- `._enums.ScoringStatistics`
+
+### Defines
+| Symbol | Type | What It Does |
+|--------|------|-------------|
+| `IndexDocumentsBatch` | class | Adds `add_upload_actions`, `add_delete_actions`, `add_merge_actions`, `add_merge_or_upload_actions`, `dequeue_actions`, `enqueue_actions`, `actions` property |
+| `RequestEntityTooLargeError` | class | `HttpResponseError` subclass for 413 |
+
+### Enum Aliases
+```python
+ScoringStatistics.Global = ScoringStatistics.GLOBAL_ENUM
+```
+
+### After Regeneration, Verify
+- [ ] `IndexDocumentsBatchGenerated` base class still exists with compatible constructor
+- [ ] `ScoringStatistics.GLOBAL_ENUM` still exists in generated enums
+
+---
+
+## File: `azure/search/documents/_operations/_patch.py`
+
+### Depends On (from generated code)
+- `..models._models.SearchRequest` (constructed directly in `_build_search_request`)
+- `..models._models.SearchResult` (field access in `_convert_search_result`)
+- `azure.core.paging.ItemPaged`, `azure.core.paging.PageIterator`
+
+### Defines
+| Symbol | Type | What It Does |
+|--------|------|-------------|
+| `_convert_search_result(result)` | function | Extracts `@search.score`, `@search.reranker_score`, `@search.highlights`, `@search.captions`, `@search.document_debug_info`, `@search.reranker_boosted_score` |
+| `_pack_continuation_token(response, api_version)` | function | Base64 JSON: `{apiVersion, nextLink, nextPageParameters}` |
+| `_unpack_continuation_token(token)` | function | Decodes token to `(next_link, next_page_request)` |
+| `_build_search_request(search_text, **kwargs)` | function | Builds `SearchRequest`. Pipe-delimited encoding for answers/captions/rewrites |
+| `SearchPageIterator` | class | Custom page iterator with `get_facets()`, `get_count()`, `get_coverage()`, `get_answers()`, `get_debug_info()` |
+| `SearchItemPaged` | class | Extends `ItemPaged` with same metadata accessors |
+| `_SearchClientOperationsMixin` | class | Overrides: `search()`, `index_documents()` (413 splitting), `upload_documents()`, `delete_documents()`, `merge_documents()`, `merge_or_upload_documents()` |
+
+### After Regeneration, Verify
+- [ ] `SearchRequest` model fields still match what `_build_search_request` sets — new search parameters need to be wired through
+- [ ] `SearchResult` model fields still match what `_convert_search_result` extracts — new `@search.*` fields need to be added
+- [ ] Generated mixin methods that `_SearchClientOperationsMixin` calls (e.g., `_search_post`) still exist with same signatures
+
+---
+
+## File: `azure/search/documents/aio/_patch.py`
+
+### Depends On (from generated code)
+- `.._client.SearchClient` as async `_SearchClient` (base class)
+
+### Defines
+| Symbol | Type | What It Does |
+|--------|------|-------------|
+| `SearchClient` | class | Async subclass, same reorder as sync |
+| `SearchIndexingBufferedSender` | class | Async version: `asyncio.Task` instead of `threading.Timer`, `iscoroutinefunction()` callback detection |
+
+### After Regeneration, Verify
+- [ ] Same checks as sync `_patch.py`
+
+---
+
+## File: `azure/search/documents/aio/_operations/_patch.py`
+
+### Depends On (from generated code)
+- Same models as sync `_operations/_patch.py`
+- `azure.core.async_paging.AsyncItemPaged`, `AsyncPageIterator`
+
+### Imports From Sync (NOT Duplicated)
+- `_build_search_request`, `_convert_search_result`, `_pack_continuation_token`, `_unpack_continuation_token`
+
+### Defines
+| Symbol | Type | What It Does |
+|--------|------|-------------|
+| `AsyncSearchPageIterator` | class | Async version of `SearchPageIterator` |
+| `AsyncSearchItemPaged` | class | Async version of `SearchItemPaged` |
+| `_SearchClientOperationsMixin` | class | Async operations mixin, same methods as sync |
+
+### After Regeneration, Verify
+- [ ] Same checks as sync `_operations/_patch.py`
+- [ ] Imports from sync `_operations/_patch.py` still resolve
+
+---
+
+## File: `azure/search/documents/indexes/models/_patch.py`
+
+### Depends On (from generated code)
+- `._models.SearchField` as `_SearchField` (base class)
+- `._models.SearchIndexerDataSourceConnection` as `_SearchIndexerDataSourceConnection` (base class)
+- `._models.KnowledgeBase` as `_KnowledgeBase` (base class)
+- `._enums.SearchFieldDataType`, `OcrSkillLanguage`, `SplitSkillLanguage`, `TextTranslationSkillLanguage`
+- Various model imports for type annotations
+
+### Defines
+| Symbol | Type | What It Does |
+|--------|------|-------------|
+| `SearchField` | class | Adds `hidden` property (inverse of `retrievable`), constructor accepts `hidden` kwarg |
+| `SearchIndexerDataSourceConnection` | class | Accepts `connection_string` str or `credentials` object |
+| `KnowledgeBase` | class | Deserializes `retrieval_reasoning_effort` from dict |
+| `SimpleField()` | function | Builder: sets `searchable=False` explicitly |
+| `SearchableField()` | function | Builder: auto-types to `String`/`Collection(String)` |
+| `ComplexField()` | function | Builder: sets type to `Complex`/`Collection(Complex)` |
+| `Collection()` | function | Wraps type in `"Collection(...)"` format |
+| `_RemovedModel` | class | Base for tombstones: raises `ValueError` on init, `TypeError` on subclass |
+| `EntityRecognitionSkill` | tombstone | "Use EntityRecognitionSkillV3 instead" |
+| `EntityRecognitionSkillLanguage` | tombstone | "Use EntityRecognitionSkillV3 instead" |
+| `SentimentSkill` | tombstone | "Use SentimentSkillV3 instead" |
+
+### Enum Aliases
+```python
+# SearchFieldDataType  (old camelCase -> generated UPPER_CASE)
+SearchFieldDataType.String = SearchFieldDataType.STRING
+SearchFieldDataType.Int32 = SearchFieldDataType.INT32
+SearchFieldDataType.Int64 = SearchFieldDataType.INT64
+SearchFieldDataType.Single = SearchFieldDataType.SINGLE
+SearchFieldDataType.Double = SearchFieldDataType.DOUBLE
+SearchFieldDataType.Boolean = SearchFieldDataType.BOOLEAN
+SearchFieldDataType.DateTimeOffset = SearchFieldDataType.DATE_TIME_OFFSET
+SearchFieldDataType.GeographyPoint = SearchFieldDataType.GEOGRAPHY_POINT
+SearchFieldDataType.ComplexType = SearchFieldDataType.COMPLEX
+
+# Monkey-patched staticmethod
+SearchFieldDataType.Collection = staticmethod(Collection)
+
+# Python keyword collisions (generated IS_ENUM -> alias IS)
+OcrSkillLanguage.IS = OcrSkillLanguage.IS_ENUM
+SplitSkillLanguage.IS = SplitSkillLanguage.IS_ENUM
+TextTranslationSkillLanguage.IS = TextTranslationSkillLanguage.IS_ENUM
+```
+
+### After Regeneration, Verify
+- [ ] All right-hand-side enum members (`STRING`, `INT32`, `IS_ENUM`, etc.) still exist
+- [ ] `_SearchField`, `_SearchIndexerDataSourceConnection`, `_KnowledgeBase` base constructors unchanged
+- [ ] No new enum values that collide with Python keywords — if so, add aliases
+- [ ] `SearchField.retrievable` property still exists (used by `hidden` inversion)
+
+---
+
+## File: `azure/search/documents/indexes/_operations/_patch.py`
+
+### Depends On (from generated code)
+- Generated `_SearchIndexClientOperationsMixin` and `_SearchIndexerClientOperationsMixin` (base classes)
+- `azure.core.match_conditions.MatchConditions`
+
+### Defines
+| Symbol | Type | What It Does |
+|--------|------|-------------|
+| `_convert_index_response(response)` | function | Maps `SearchIndexResponse` → `SearchIndex` (semantic → semantic_search) |
+
+**`_SearchIndexClientOperationsMixin`** wraps:
+| Method | Customization |
+|--------|--------------|
+| `delete_index(index)` | Polymorphic: str or SearchIndex |
+| `create_or_update_index(index)` | Adds prefer, match_condition, allow_index_downtime |
+| `delete_synonym_map(synonym_map)` | Polymorphic: str or SynonymMap |
+| `create_or_update_synonym_map(synonym_map)` | Adds prefer, match_condition |
+| `delete_alias(alias)` | Polymorphic: str or SearchAlias |
+| `create_or_update_alias(alias)` | Adds prefer, match_condition |
+| `delete_knowledge_base(kb)` | Polymorphic: str or KnowledgeBase |
+| `create_or_update_knowledge_base(kb)` | Adds prefer, match_condition |
+| `delete_knowledge_source(ks)` | Polymorphic: str or KnowledgeSource |
+| `create_or_update_knowledge_source(ks)` | Adds prefer, match_condition |
+| `list_indexes(*, select)` | Uses `_convert_index_response` for projections |
+| `list_index_names()` | Name-only projection via `cls` callback |
+| `get_synonym_maps(*, select)` | Returns list |
+
+**`_SearchIndexerClientOperationsMixin`** wraps:
+| Method | Customization |
+|--------|--------------|
+| `delete_data_source_connection(dsc)` | Polymorphic: str or object |
+| `create_or_update_data_source_connection(dsc)` | Adds prefer, match_condition, skip_indexer_reset |
+| `delete_indexer(indexer)` | Polymorphic: str or SearchIndexer |
+| `create_or_update_indexer(indexer)` | Adds prefer, match_condition, skip/disable cache |
+| `delete_skillset(skillset)` | Polymorphic: str or SearchIndexerSkillset |
+| `create_or_update_skillset(skillset)` | Adds prefer, match_condition, skip/disable cache |
+
+### After Regeneration, Verify
+- [ ] All generated `_delete_*`, `_create_or_update_*`, `_list_*` base methods still exist with compatible signatures
+- [ ] `SearchIndexResponse` still has `.semantic` field (used by `_convert_index_response`)
+- [ ] New resource types added to spec → add polymorphic delete/update wrappers here
+- [ ] Verify prefer header value `"return=representation"` still applies
+
+---
+
+## File: `azure/search/documents/indexes/aio/_operations/_patch.py`
+
+Async mirror of `indexes/_operations/_patch.py`. Same methods, all `async`.
+
+Imports `_convert_index_response` from sync — NOT duplicated.
+
+### After Regeneration, Verify
+- [ ] Same checks as sync `indexes/_operations/_patch.py`
+- [ ] Import of `_convert_index_response` from sync still resolves
+
+---
+
+## Empty `_patch.py` Files (No Customizations)
+
+These contain only `__all__ = []` and empty `patch_sdk()`. No action needed after regeneration.
+
+- `azure/search/documents/indexes/_patch.py`
+- `azure/search/documents/indexes/aio/_patch.py`
+- `azure/search/documents/knowledgebases/_patch.py`
+- `azure/search/documents/knowledgebases/models/_patch.py`
+- `azure/search/documents/knowledgebases/aio/_patch.py`