HISTORY.md

## 11.7.0 (November 3, 2026)

### New Features

- **Eight new connector types**: `AtlanConnectorType` gains `DREMIO`, `SSRS`, `MONGO`, `ANOMALO`, `AZURE_DATA_FACTORY`, `SAP_ANALYTICS_CLOUD`, `SAP_DATASPHERE` and `SAP_S4`, the names the matching connector apps already write, so `Connection.creator()` no longer needs `CREATE_CUSTOM` for them.

### Experimental: pyatlan_v9

- **`creator()` for 43 more asset types across 18 connectors**: SageMaker (5), SSRS (4), SAP Analytics Cloud (4), Unstructured/bridge (4), ADF (3), Dremio (3), DynamoDB (3), SAP ERP (3), Kafka and Schema Registry (3), MongoDB (2), Monte Carlo (2), and one each for `Function`, PowerCenter `Flow`, `MatillionComponent`, `FivetranConnector`, `AnomaloCheck`, `SnowflakeShare` and `SapDatasphereReplicationFlow`. Each creator reproduces the qualifiedName its connector app already emits, byte for byte, so no asset changes identity, and each comes with `updater()` / `trim_to_required()`, a tightened `_QUALIFIED_NAME_PATTERN` and per-type tests. `Function.creator()` takes a schema of any depth through the FND-3887 parent resolution, and `FivetranConnector` builds its hashed id and URLs exactly as the app does.

### QOL Improvements

- **pydantic advisories cleared on Python 3.10+**: The `pydantic` pin is split by interpreter: `~=2.14.2` on Python 3.10 and newer (clears GHSA-49g6-96jj-f7r9, GHSA-955g-x97g-8mcp, GHSA-jfv7-ghgv-gx4x and GHSA-qpmf-fgh3-ff25), `~=2.13.4` kept on 3.9, which the fixed release does not support. Installing pyatlan on 3.10+ now pulls pydantic 2.14.2 or later.
- **CI**: `pyatlan-pr` runners are pinned to `ubuntu-24.04` (`ubuntu-latest` moved to 26.04, which has no Python 3.9 build), and the new authlib advisories are ignored on the Python 3.9 scans only, following the anyio precedent; 3.10+ already resolves the fixed authlib.

## 11.6.0 (October 26, 2026)

### New Features

- **Cloud SQL for PostgreSQL and AlloyDB app builders (AICHAT-2025)**: `pyatlan.model.apps` gains `AtlanCloudsqlPostgresCrawler` (basic, IAM user, IAM service account and Workload Identity Federation auth), `CloudsqlPostgresMiner` (mines an existing connection's query history) and `AtlanAlloydbPostgres` (basic, IAM user and IAM service account auth), generated from the apps' UI configmaps. Each carries `redact_sql_text()`, and the existing `load(slug).update()` path turns Redact SQL Text on for workflows that pre-date the toggle, keeping the connection and credential reference. The generator now encodes a boolean widget over a string contract as the `"true"`/`"false"` the server expects.
- **Search scoped to the caller's full access**: `IndexSearchRequest` gains `scope_to_user_access` (`scopeToUserAccess`), alongside `enable_full_restriction`. When `True`, the server returns the union of assets the caller's enabled personas and purposes grant, in one request; admins stay unscoped. Unset by default, so existing requests are unchanged.
- **Latest `pyatlan_v9` models**: Resynced from `atlanhq/models`, adding `SAPDatasphereReplicationFlow` and the SAP component, data type, field count and order, logical name, package name and technical name attributes on the SAP related types. Nothing is removed.

## 11.5.0 (October 22, 2026)

### New Features

- **`GlueTag` and the `TagAttachment` source tag contract (CONNECT-2796)**: Adds the `GlueTag` asset (supertypes `Tag`, `SQL`) with `glueTagType` and the `GlueTagType` enum (`property`, `resource_tag`, `lf_tag`) to both `pyatlan.model.assets` and `pyatlan_v9`, and adds `TagAttachment` to `pyatlan_v9`. In both packages `TagAttachment` now carries the eight source tag fields publish-app reads (`objectQualifiedName`, `objectTypeName`, `sourceTagQualifiedName`, `sourceTagDisplayName`, `sourceTagTypeName`, `valueType`, `value`, `propagate`), so connectors map tags through a model instead of hand-written dicts.
- **New connector types**: `AtlanConnectorType.AWS_SMUS` (`aws-smus`, SaaS) for AWS SageMaker Unified Studio, so `Connection.creator()` no longer needs `CREATE_CUSTOM` for it; plus `FABRIC`, `SAGEMAKER` and the Cosmos MongoDB connector types.
- **Data product lineage (AICHAT-1984)**: `LineageListRequest` and `FluentLineage` gain `lineage_type`, with the `LineageType` enum (`DATASET_PROCESS_LINEAGE`, `PRODUCT_ASSET_LINEAGE`). Lineage for a `DataProduct` came back empty because the default lineage type treats a product as an unconnected node; `PRODUCT_ASSET_LINEAGE` returns what the product Lineage tab shows. Unset by default, so existing requests are unchanged. Same in `pyatlan_v9`.
- **Latest asset models (pydantic + `pyatlan_v9`)**: Both model sets are regenerated from the latest typedefs. `pyatlan_v9` gains `AzureEventHubConsumerGroup`, `BigqueryTag`, `DatabricksUnityCatalogTag`, `ConnectionProcess`, `QlikStream`, the DynamoDB secondary index types and the business policy exception, incident and log types; the pydantic models pick up the latest attributes across the Cosmos MongoDB, Databricks, Snowflake AI model, Fabric and Power BI types. Nothing is removed.

### Experimental: pyatlan_v9

- **`creator()` for more than 130 asset types across 23 connectors**: Cognos (10), Cosmos MongoDB (3), Iceberg (4), Salesforce (5), QuickSight analysis field, Sigma (7), Soda check, SAP HANA calculation view, MicroStrategy (10), Looker (9), Tableau (10), dbt (11), Microsoft Fabric (11), Power BI (12), Databricks, Qlik Sense (7), Cassandra (5), Dataplex (2), Redash (3), Domo (4), ThoughtSpot (6), Mode (5), and the shared `TagAttachment`, `S3Prefix` and `FlowControlOperation` types. Each creator reproduces the qualifiedName its connector app already builds, so no asset changes identity, and each comes with `updater()` / `trim_to_required()` and per-type tests.
- **SQL creators accept non-standard depths and nested columns (FND-3887)**: With `connection_qualified_name` given, `Table`, `View`, `MaterialisedView`, `Procedure`, `Column` and `Schema` accept parents of any depth (Fabric's six segments, Db2 z/OS's four). `Table` can be schema-less and `Schema` database-less, so `schema_qualified_name` / `database_qualified_name` are now optional keywords. `Column.creator` takes `parent_column=` for nested columns (setting depth, hierarchy and the parent relationship) and accepts `MongoDBCollection` and `DocumentDBCollection` parents. Behaviour change: `Column.creator(parent_type=Column)` without `parent_column` now raises "parent_column is required when parent_type is Column".
- **Process creators are stricter and deterministic (FND-3888)**: `Process`, `ColumnProcess` and `BIProcess` raise `ValueError` on an empty or whitespace `process_id` instead of silently falling back to the hash identity (`None` still hashes). `extra_hash_params` is read in sorted order, so a hash over two or more params no longer depends on `PYTHONHASHSEED`.
- **Model overlays are lint-clean once injected (FND-3933)**: The hand-written overlays now match what the repo's pre-commit hooks produce, so model sync PRs carry only real model changes.

### QOL Improvements

- **Dependencies refreshed**: authlib 1.6.12 (Python 3.9) / 1.9.1 (3.10+), virtualenv 21.14.7, certifi 2026.10.19, idna 3.15, and the rest of the lock.
- **CI**: The Trivy `uv.lock` gate ignores anyio CVE-2026-63374 for the Python 3.9 lock entry only, the same way the pip-audit step already did. The fixed anyio needs Python 3.10, which 3.10+ already resolves; 3.9 stays supported.

## 11.4.1 (October 9, 2026)

### New Features

- **Latest asset models (pydantic + `pyatlan_v9`)**: Both model sets are regenerated from the latest typedefs. The pydantic models (`pyatlan.model.assets`) gain the SAP Analytics Cloud types (models, stories, folders, columns), `KafkaTag` and `FabricSemanticModelMeasure`, plus the latest attributes across existing connectors. Both sets gain `assetManagedBy` — the identity of the agent that creates and maintains an asset — and `pyatlan_v9` also gains `knowledgeLinkedFiles` on the glossary types. Nothing is removed.
- **`Notification` asset model**: Adds the `Notification` asset type and its related enums.
- **Kafka Confluent app gains Cloud API, Cloud Metrics and Connect Lineage**: The `KafkaConfluent` builder takes `enable_cloud_api` on `basic()` and the `include_cloud_metrics` / `include_connect_lineage` workflow toggles, matching what the app surfaces.
- **`BIProcess.creator()` in `pyatlan_v9`**: `BIProcess` gains `creator()`, `updater()`, `trim_to_required()` and `generate_qualified_name()`, so BI connectors no longer hand-build the process qualifiedName. The grammar matches `Process`: an explicit `process_id` gives `{connection_qualified_name}/{process_id}`, otherwise the same deterministic hash of inputs and outputs. Clears the application-sdk conformance rule P028 for BI lineage.

### Bug Fixes

- **`BIProcess.creator()` returns a `BIProcess`**: In `pyatlan.model.assets`, `BIProcess.creator()` (and the deprecated `create()`) inherited `Process.creator` and returned a `Process` with `type_name == "Process"`. It now returns a `BIProcess`, building the qualifiedName the same way `ColumnProcess` does.

## 11.4.0 (September 18, 2026)

### New Features

- **Latest asset models (pydantic + `pyatlan_v9`)**: Regenerated the pydantic (`pyatlan.model.assets`) and experimental `pyatlan_v9` (msgspec) asset models from the latest Atlan typedefs. Adds new asset types — SAP Analytics Cloud (models, stories, folders, columns), `KafkaTag`, and `FabricSemanticModelMeasure` — plus the latest attributes across existing connectors.

## 11.3.0 (September 17, 2026)

### New Features

- **Link knowledge files to assets**: `AssetClient` (sync and async) gains `append_knowledge_files`, `replace_knowledge_files` and `remove_knowledge_files`, so a `KnowledgeFile` can be linked to any asset the same way glossary terms are — append and remove are idempotent, and replace sends the full set (`[]` unlinks every file).

### Experimental: pyatlan_v9

- **`to_atlas_format` preserves explicitly-set nulls**: Clearing an attribute with `asset.description = None` now emits a JSON null instead of being silently dropped, so the clear reaches Atlas. `UNSET` fields stay absent, as before. This changes `to_atlas_format` output for every caller.

## 11.2.0 (September 16, 2026)

### New Features

- **Latest asset models (pydantic + `pyatlan_v9`)**: Both the pydantic (`pyatlan.model.assets`) and experimental `pyatlan_v9` (msgspec) asset models are regenerated from the latest Atlan typedefs. Adds the newest asset types — including `DatabricksGenieAgent` — and the latest attributes across connectors, and promotes `KnowledgeFile` / `KnowledgeFolder` (with their `Knowledge` base) into the core asset set.

### Breaking Changes

- **Removed platform-deprecated asset types**: Dropped types that no longer exist in the Atlan typedefs — the `Coalesce` / `V1Coalesce` family, the Snowflake V1 Cortex types (`SnowflakeV1CortexAgent` and related), `KafkaPartition`, and `SigmaDataModelElement` — along with their relationship attributes. Code importing these classes directly will need to be updated.

### Experimental: pyatlan_v9

- **Related decode preserves the wire typeName (AICHAT-1883)**: `Related{Type}.__post_init__` no longer overwrites the `typeName` decoded from the wire with the class default. A relationship that arrives as a concrete subtype (e.g. `SigmaDataElementField` decoded into a broader `RelatedSigma` field) now keeps its real type instead of collapsing to the parent's default.
- **Lifecycle methods on regenerated assets**: The regenerated `pyatlan_v9` assets emit the per-asset `validate()` / `minimize()` / `relate()` lifecycle methods.
- **Lineage fields preserved in nested conversion (BLDX-711)**: Restored copying of `depth` / `immediate_upstream` / `immediate_downstream` through the nested `_from_nested` conversion, so lineage-API fields survive round-tripping.
- **`typeName` always serializes (BLDX-710)**: Dropped the redundant per-class `type_name` field default. Because the msgspec structs use `omit_defaults=True`, a value equal to its own class default was being omitted from the wire; `type_name` is now declared once (as `UNSET`) so `typeName` always serializes.

### QOL Improvements

- **CI**: Removed the automated Claude Code PR-review workflow.

## 11.1.0 (September 10, 2026)

### Bug Fixes

- **Miner runs no longer rename the connection they mine (AICHAT-1798)**: A miner created by referencing an existing connection *only* by `qualified_name` (e.g. `BigqueryMiner(client).connection(qualified_name=...).run()`) sent a connection object with no `name`. The downstream popularity/publish step derives the connection name from that payload and, finding it empty, fell back to the qualifiedName's numeric tail — renaming the connection to that number. A fresh miner run now reads the connection's stored config back (identity, credential, admins, query settings — the same shape the UI sends, without internal analytics/popularity fields) and sends it with the name intact, so the connection cannot be renamed. If the connection cannot be read and no explicit `connection(name=...)` was given, the run now raises `CONNECTION_READ_FOR_APP_FAILED` rather than silently renaming it. Crawler `create`, `load()` / `update()`, and rerun (`submit`) are unaffected.

## 11.0.1 (September 9, 2026)

### Packages

- **pyatlan container image also published to GHCR**: The pyatlan image now mirrors its full Harbor tag ladder to `ghcr.io/atlanhq/pyatlan`, so it can be pulled from GitHub Container Registry as well as Harbor. CI/distribution only — the Python SDK is unchanged from 11.0.0.

## 11.0.0 (August 26, 2026)

### New Features

- **`load()` / `update()` for app workflows**: Every app builder now supports `load(slug)` + `update()` to publish a new version of an existing workflow, changing only the fields you set while preserving everything else — including reusing the persisted credential by guid (never re-vaulted or rotated). Because it lives on the base `AppBuilder`, every connector builder (BigQuery, Snowflake, …) inherits the same fluent update path as create.
- **Typed Standard Lineage builder (`AtlanStandardLineage`)**: A new builder in `pyatlan.model.apps` for building lineage across multiple connections of the same connector, reading each connection's already-mined query history (no crawl, no credential). Includes re-scope helpers (`add_connections` / `remove_connections` / `set_connections` / `get_connections`) for onboarding connections over time. Generated from the app's input contract via a reusable overlay so it stays in sync as the contract evolves.
- **Idempotent app `submit`**: `client.app.submit(slug, idempotent=True)` refuses to launch a duplicate run while one is already in progress, raising `APP_WORKFLOW_ALREADY_RUNNING` instead of starting another. Resilient to search-index lag (it reconfirms against the index if the server rejects the submit). Defaults to `False`, so existing callers are unaffected. Sync and async parity.
- **Asset Export (Basic) "Export empty custom metadata" toggle**: `CsaUberAssetExportBasic` now surfaces `export_empty_custom_metadata("true" | "false")`. When on (the default), the export emits a column for every custom metadata attribute defined on the tenant — even when no exported asset has a value for it — restoring the legacy behaviour where defined-but-unfilled attributes still produced columns; off emits only attributes valued somewhere in the export.
- **BigQuery miner billing strategy**: `BigqueryMiner` now carries `billing-strategy` — `central` (default, today's behaviour) or `distributed`, which bills each mined project for its own query-history jobs instead of the connection's credential project.

### Breaking Changes

- **Removed the legacy Argo workflow client and package builders**: `client.workflow` (`WorkflowClient` / `AsyncWorkflowClient`, along with its shared request helpers, workflow models, and workflow API constants) and the `pyatlan.model.packages.*` builders have been removed. They only existed to orchestrate the legacy Argo workflow surface, which no longer runs on native-app tenants. Use `client.app` (see [Manage apps](https://docs.atlan.com/product/capabilities/build-apps/sdks/python/apps/manage-apps)) and the per-connector builders in `pyatlan.model.apps` instead. The `Workflow` **asset** entity and the approval-workflow client (`client.inbox`) are unaffected.
- **Typed BigQuery WIF credentials**: `BigqueryCrawler.workload_identity_federation()` now takes the four WIF-specific values — `service_account_email`, `wif_pool_provider_id`, `atlan_oauth_id`, and `atlan_oauth_secret` — as typed, required keyword-only parameters, matching the fully-typed `service_account()`. Previously these rode through untyped `**extra`, so they were undiscoverable and a misspelled key was silently accepted (producing a broken `gcp-wif` credential with no error). **Impact:** a bare `project_id`-only call now raises `TypeError`; pass the WIF values explicitly. `**extra` is retained for forward-compatibility with fields newer than this signature.

### Experimental: pyatlan_v9

- **Models regenerated to the latest typedefs**: `pyatlan_v9` (msgspec) is regenerated from the latest type definitions ([models#1899](https://github.com/atlanhq/models/pull/1899)), adding new asset families (AI, Snowflake Cortex) and the newest attributes. This regeneration also **restores the per-asset `validate()` / `minimize()` / `relate()` methods** on every asset (a regression guard now asserts they stay present), and normalizes import ordering across the tree.

### Bug Fixes

- **App-workflow `update()` preserves the full connection**: `update()` now re-sends the complete persisted connection and re-injects `connection_qualified_name` (which the server derives on create but not on update), so a run submitted after an update no longer fails downstream (for example, Publish-to-Atlas). String-typed contract fields returned as objects are normalized, and every other current input is preserved.
- **Reused credential placement (CONNECT-843)**: When reusing an already-vaulted credential, the guid now rides on the connection entity (`attributes.defaultCredentialGuid`) with top-level `credential_guid` left empty — the UI's wire shape — instead of at the top level. A top-level guid with no credential body made the create/update endpoint flatten the shared credential record (wiping a `gcp-wif` credential's `authType`/`extra`), breaking every workflow that shared it. `load()` reads the reused guid back from the connection (falling back to top-level for older workflows).
- **App management no longer retries HTTP 500**: App-management calls no longer retry on HTTP 500. A 500 there is not transient (for example, submitting a workflow that already has an active run), and retrying a non-idempotent POST could trigger multiple failed runs; a single `submit` is now exactly one attempt.
- **App-workflow `update()` now applies builder changes for every connector**: `load(slug).<setter>().update()` previously only took effect for the few builders whose setter keys happen to be underscored (e.g. `BigqueryCrawler`); for the ~32 builders that key off the configmap's hyphenated field ids (`BigqueryMiner`, `SnowflakeCrawler`, the `Atlan*` connectors, …) the change was silently dropped, because `load()` reads the workflow's underscore arg keys and the loaded value won the collision. `update()` now collapses each input key to its underscore field-name form (verified: every field name equals its alias with hyphens→underscores across all 35 input models), so a builder-set value always overrides the loaded one. Runtime/structural keys (`user-id`, `connection`, `credential`) are unaffected.
- **Clearer "already running" submit error**: When `submit(..., idempotent=True)` refuses a duplicate run, the error now explains how to proceed — wait for the current run to finish, cancel it (`client.app.cancel_run(run_id)` or from the Atlan UI), or pass `idempotent=False` to submit anyway — and links the [Manage apps](https://docs.atlan.com/product/capabilities/build-apps/sdks/python/apps/manage-apps) docs. It no longer surfaces the internal automation-engine run identifier.

### QOL Improvements

- **pydantic assets synced to the latest typedefs**: The `pyatlan.model.assets` (pydantic) set is regenerated against the latest type definitions, picking up new asset types and attributes, with several generator fixes (deterministic cycle-safe imports, correct core/non-core import paths, complete struct imports) and import ordering normalized.

### Packages

- **cryptography**: Bumped from 49.0.0 to 50.0.0 to address CVE-2026-69247 (HIGH).

## 10.0.0 (August 13, 2026)

### New Features

- **Manage requests and approvals programmatically (`client.requests` and `client.inbox`)**: Two new clients cover both of Atlan's approval systems. `client.requests` lists, creates, approves, and rejects requests in the classic Requests module (Governance Center → Requests) with typed filters, enums (`AtlanRequestStatus`, `AtlanRequestType`), and lazy pagination. `client.inbox` inspects and bulk-approves or bulk-rejects governance-workflow tasks (the Inbox) — the first SDK to support it — with group-scoped `approve_all` / `reject_all` keyed by a task GUID (one task) or an asset GUID (all pending tasks on it). Sync and async parity throughout.
- **Asset Export (Basic) v3 app builder (`CsaUberAssetExportBasic`)**: A new typed builder in `pyatlan.model.apps` runs Asset Export on v3 native-app tenants, where the legacy Argo-template path (`packages.AssetExportBasic` + `client.workflow.run()`) no longer resolves. Supports `DIRECT` / `EMAIL` / `CLOUD` delivery and `s3()` / `gcs()` / `adls()` object-store targets, mirroring the legacy builder's ergonomics. Every legacy `model.packages` class that has a typed v3 builder now emits a deprecation warning naming its replacement.

### Breaking Changes

- **`AtlanTag` now sends its propagation defaults on the wire**: Previously, constructing a tag as `AtlanTag(type_name=t)` sent no `propagate` field, so the server applied its own default (`propagate=True`) — the SDK documented `False`, but such tags propagated anyway. A new `AtlanObjectWithDefaults` base marks fields with a non-`None` declared default as set at construction, so a bare `AtlanTag` now serializes all four propagation flags with their documented defaults (`propagate=false`, `removePropagationsOnEntityDelete=true`, `restrictPropagationThroughLineage=false`, `restrictPropagationThroughHierarchy=false`). **Impact:** tags created without explicit propagation settings will no longer propagate by default. Set the flags explicitly if you relied on the previous server-side behavior.

### Documentation

- **Manage requests**: [Managing requests with the Python SDK](https://docs.atlan.com/product/capabilities/build-apps/sdks/python/how-tos/manage-requests) — `client.requests` (classic Requests module).
- **Manage governance workflow approvals**: [Managing the Inbox with the Python SDK](https://docs.atlan.com/product/capabilities/build-apps/sdks/python/how-tos/manage-governance-workflow-approvals) — `client.inbox`.
- **Run the Asset Export Basic app**: [Asset Export Basic with the Python SDK](https://docs.atlan.com/product/capabilities/build-apps/sdks/python/apps/utilities/asset-export-basic) — `CsaUberAssetExportBasic`.

## 9.11.0 (July 29, 2026)

### New Features

- **`SqlInsightJoin.creator()` with miner-identical qualifiedName**: Added a typed `creator()` for `SqlInsightJoin` (previously only `updater()` existed), so SDK callers — the Atlan MCP server's `create_sql_insight_join` tool first among them — can author human-confirmed SQL-Intelligence join relationships without hand-building raw Atlas payloads. The new public `SqlInsightJoin.generate_qualified_name()` derives the deterministic identity (`source_qn/join/md5(joined_qn|sorted_pairs|join_type)`) byte-identically to the SQL-Intelligence miner, so a human-confirmed join and a later mined observation of the same join converge on one entity instead of duplicating. The creator writes both anchorings — the string qualified-name attributes and `sqlInsightJoinColumnPairs` (read by metadata-lakehouse consumers) and both `sqlInsightSourceDataset`/`sqlInsightJoinedDataset` relationship edges (rendered on the asset page) — and pins `queryCount`/`uniqueUsers` to 0, since a human-declared join has no observed usage. Covered by unit tests (including a byte-exact digest golden against a live mined row) and integration tests (create, read-back with live edges, upsert convergence on the same guid).

## 9.10.3 (July 22, 2026)

### New Features

- **Metabase `creator()` factories**: Added `creator()` factory methods for the Metabase asset family — `MetabaseCollection` (`name`, `connection_qualified_name`, `metabase_id`), `MetabaseDashboard`, and `MetabaseQuestion` — so the qualifiedName grammar is owned centrally by the SDK instead of connectors hand-building qualifiedNames with f-strings (resolves the application-sdk `P028 ManualQualifiedNameFString` conformance warnings). Also available in the experimental `pyatlan_v9` package, with integration tests covering the new factories.

## 9.10.2 (July 21, 2026)

### New Features

- **New `INFORMATICA_POWERCENTER` connector type**: Added `INFORMATICA_POWERCENTER` to the supported connector types. (PARTCON-28)

### Packages

- **Dependencies refreshed while keeping Python 3.9 support**: Runtime and dev dependencies were bumped to their current releases using `python_version` marker splits, so packages that have dropped Python 3.9 stay pinned to their last 3.9-compatible version on 3.9 and move forward on 3.10+. Includes `httpx-retries`, `authlib`, `tenacity`, and `networkx`, plus dev tooling (`pytest` 9, `mypy` 2.3, `vcrpy` 8, `deepdiff` 9, and others). No API changes; verified across Python 3.9 and 3.13. (BLDX-1571)
- **`msgspec` cap lifted on Python 3.10+**: `msgspec` now resolves to `~=0.21.1` on Python 3.10+ to avoid a known Python 3.13 concurrent-decode segfault, while Python 3.9 stays on `~=0.20.0`. (CNCT-85)

## 9.10.1 (July 15, 2026)

### Bug Fixes

- **Bulk audit search no longer fails on large result sets**: Iterating a bulk audit search (`client.audit.search(..., bulk=True)`) over more than 10,000 matching events could fail partway with an Elasticsearch "Result window is too large" error and stop without returning the rest. Pagination now stays within the result-window limit no matter how many events match, so large audit scans complete and return every record.

## 9.10.0 (July 14, 2026)

### Bug Fixes

- **Clearer error when adding a tag that doesn't exist**: Adding an Atlan tag by a name that isn't in your workspace now fails right away with a clear message that names the tag (`Atlan tag with name <name> does not exist.`), instead of a confusing server error that made it look like the tag had been deleted. Reading and re-saving assets that already carry a deleted tag continues to work as before.
- **Much faster imports**: Importing asset models (for example, `from pyatlan.model.assets import Column`) is now about 3× faster — a cold import drops from roughly 8 seconds to under 3. The improvement is largest on Windows, where imports could previously stall. Everything else behaves exactly the same.

### Packages

- **cryptography upgraded** to pick up an important security fix. Note: this now requires Python 3.9.1 or newer.

## 9.9.0 (June 30, 2026)

### New Features

- **Python 3.14 support**: pyatlan now supports Python 3.14. The model layer's pydantic v1 compatibility shim was incompatible with Python 3.14's deferred-annotation semantics (PEP 649 / PEP 749) on pydantic 2.12.x — `AtlanClient` failed to import with `ConfigError: unable to infer type for attribute ...`. Bumping pydantic to 2.13.x restores the v1 compat layer on 3.14; no API changes. Verified across Python 3.9–3.14. (#962)
- **New policy-action enums (`PersonaAIAction`, `AdminAction`, `TypeDefAction`)**: Added to `pyatlan.model.enums`, mirroring the Java SDK and matching the shape of the existing `PersonaMetadataAction` / `DataAction`. `PersonaAIAction` covers the 18 `persona-ai-application-*` / `persona-ai-model-*` Ranger slugs; `AdminAction` and `TypeDefAction` cover the `admin-*` and `type-*` / relationship slugs — so tooling can convert raw Ranger policy-action slugs to their canonical enum names. (#961)

### Bug Fixes

- **App create — idempotent reuse on a bare `409` duplicate-name**: `client.app.create(...)` now reliably reuses an existing workflow on a duplicate name. The create endpoint's `409` body carries no `code`/`status` field, so it surfaced as a bare `AtlanError` rather than `ConflictError` and the create-or-reuse path never triggered; `create` (sync + async) now handles that case, reading the existing slug straight from the response body (with a by-name lookup fallback) and re-raising only when the name is non-unique. (#960)

### Packages

- **pydantic** bumped from `~=2.12.5` to `~=2.13.4` (enables Python 3.14 support). (#962)

## 9.8.0 (June 24, 2026)

### New Features

- **App workflow client (`client.app`)**: A new `AppClient` (sync + async) for creating, running, scheduling, and managing app workflows over the `/v1/app*` APIs. Workflows are created from an `app_id` plus a generic `inputs` dict validated server-side against the app's live input contract (`client.app.get_input_contract(...)`), so a connector never needs a hand-maintained package class. Includes `describe`, `get_input_contract`, `create`, `get`, `get_all`, `update`, `submit`, `get_run`, `cancel_run`, `add_schedule`, `remove_schedule`, and `delete`. (#959)
- **Typed app builders (`pyatlan.model.apps`)**: A fluent, UI-equivalent builder per connector app (e.g. `BigqueryCrawler`, `SnowflakeCrawler`, `AtlanTableau`, `PostgresCrawler`, …) mirroring the "new app" wizard (Credential → Connection → Metadata). A builder vaults the credential, mints the connection qualified name, and assembles the full payload — so callers never hand-build a connection object or guess an input key, e.g. `BigqueryCrawler(client).service_account(...).connection(name=...).include({...}).run()`. Each builder and its typed `*Inputs` model are generated from the app's UI configmaps via `uv run generate-apps`; 33 connectors are covered. (#959)
- **Resolve a workflow by name + idempotent create**: `client.app.get_all(name=...)` now filters by exact workflow name (server `?name=`), so a stateless script can resolve a slug from a name. `client.app.create(...)` is now idempotent on a duplicate name — on the server's `409` it resolves and returns the existing workflow (*create-or-reuse-by-name*) instead of raising, so re-running a migration script is safe. (#959)

### Bug Fixes

- **App builders — correct per-connector filter shapes**: BI/object filters for Tableau, Power BI, QuickSight, and Sigma are now serialized to the JSON-string form their input contract expects (`{id: {}}` → a JSON string), and Power BI's `odbc_dsn_config_mapping` is sent as an object — fixing `1000 inputs do not match the app input contract` errors. (#959)
- **Miner `extraction_method`**: the Oracle, Postgres, and Teradata miners now send `extraction_method="query_history"` (per their configmaps) instead of the default `"direct"`; the Oracle miner's `start_date` is documented as an epoch timestamp. (#959)
- **Entrypoint fallback on create**: an app that declares a named entrypoint (e.g. `crawler`) but registers its input contract only at the default slot no longer fails with `1003`; `create` retries without the entrypoint so the server resolves the default contract. (#959)

### Deprecations

- **`WorkflowClient` (`client.workflow`) and `model.packages.*`** are deprecated and now emit a `DeprecationWarning`. They target the legacy workflow surface, which no longer runs on newer Atlan tenants — use `client.app` instead. (#959)

## 9.7.6 (June 16, 2026)

### Bug Fixes

- **`Batch`/`AsyncBatch` tracked lists now carry real identity (both `pyatlan` and `pyatlan_v9`)**: the tracked lists (`created` / `updated` / `partial_updated` / `restored`) now carry each asset's real `guid` and `qualified_name`, so callers can match tracked items back to their inputs by either key. Previously non-glossary assets got a placeholder (negative) guid in `pyatlan` (or a missing/`UNSET` guid in `pyatlan_v9`) via `trim_to_required()`, and glossary terms got the guid in place of `qualified_name` in `pyatlan` (or a missing/`UNSET` `qualified_name` in `pyatlan_v9`) via `ref_by_guid()`, making the tracked items impossible to match back by guid or qualified_name. Tracking also now handles `AtlasGlossaryCategory` (and types it correctly as a category): previously a category in the tracked path crashed `__track` with `ValueError("anchor.guid must be available")`, because — like `AtlasGlossaryTerm` — its `trim_to_required()` requires the parent glossary anchor, which is absent in an `AssetMutationResponse`. (#953)

## 9.7.5 (June 16, 2026)

### Bug Fixes

- **Count partial updates in `Batch`/`AsyncBatch`**: Successful *partial* updates — primitive-only attribute changes (e.g. `user_description` alone) and any change touching a relationship (e.g. a glossary term's `anchor`) — were never counted: they reported `num_updated: 0` and were mislabeled as `restored`, causing retry storms for callers that retry on "nothing updated" (notably the common glossary-term path). They are now reconciled from both `mutated_entities.PARTIAL_UPDATE` and the top-level `partial_updated_entities` (de-duplicated by GUID) into new `partial_updated` / `num_partial_updated` trackers. (#951)
- **Fix `MutatedEntities.PARTIAL_UPDATE` alias collision**: the field was mistakenly declared with `alias="DELETE"`, colliding with the real `DELETE` field — so wire `PARTIAL_UPDATE` payloads were dropped and deleted assets leaked into `assets_partially_updated()`. Corrected to `alias="PARTIAL_UPDATE"`. (#951)

## 9.7.4 (June 2, 2026)

### Experimental: `pyatlan_v9`

- Generated latest typedef models. (#947)

## 9.7.3 (May 25, 2026)

### New Features

- **`skillArtifactContent` attribute on `SkillArtifact`**: Adds the `skill_artifact_content` (`skillArtifactContent`) field to `SkillArtifact` in both `pyatlan` and `pyatlan_v9`. The Atlas API returns this attribute on SkillArtifact entities but PyAtlan was silently dropping it. This unblocks surfacing skill `.md` body content via `get_asset` / `search_assets` without needing a raw HTTP fallback. (#937)

### Bug Fixes

- **`Connection.creator` validates connector_type slug (BLDX-1294)**: `Connection.creator` and `Connection.creator_async` (both pyatlan and pyatlan_v9) now validate `connector_type.value` against `^[a-z0-9-]+$` — the same pattern the Atlan platform's server-side asset-import (RAB) path enforces. Raises `InvalidRequestError` (`ATLAN-PYTHON-400-079`) on violation. Mirrors the Java SDK constraint. Root cause: a customer-created `dev_cmdr` connection (underscore in slug) caused RAB to reject every asset-import referencing it, leaving phantom `Connection` rows in Atlas. (#939)

## 9.7.2 (May 18, 2026)

### New Features

- **Add `AMAZON_MSK` to `AtlanConnectorType`**: New `AMAZON_MSK = ("amazon-msk", AtlanConnectionCategory.EVENT_BUS)` enum entry, matching the Java SDK. Removes the need for the `extend_enum` workaround. (PR #928)

### Bug Fixes

- **`DataProduct.creator` now defaults `daap_visibility` to `PRIVATE`**: Mirrors the Atlan UI default. Fixes the marketplace **Overview Assets** tile rendering blank for products created via the SDK because the UI required both `daapStatus` and `daapVisibility` to be non-null. The earlier partial fix in PR #508 only set `daap_status`. New optional overrides also exposed on `creator()` / `create()`: `daap_visibility`, `daap_visibility_users`, `daap_visibility_groups`, `owner_users`, `owner_groups`. (PR #932)
- **`DataProduct.creator` defaults `owner_users` to the calling user**: When a client is passed, the creating user's username is used as the default owner, matching UI behavior. (PR #932)

## 9.7.1 (May 1, 2026)

### Bug Fixes

- **`DbtProcess` and `DbtColumnProcess` now extend the correct process parent**: Atlas typedefs list `Dbt` first in `superTypes`, causing the class generator to pick `Dbt` over `Process`/`ColumnProcess` as the Python superclass. This made `issubclass(cls, Process)` return `False`, so `atlan-publish-app` placed dbt-process types in Batch 1 alongside their upstream entities, producing ATLAS-404 errors on every `inputs`/`outputs` lookup (~35 publish failures per live dbt run). Fixed via `_SUPERCLASS_OVERRIDES` — the same mechanism used for `DbtMeasure` in 9.7.0. (PR #923)

## 9.7.0 (April 30, 2026)

### New Features

- **`GENERIC_OPENLINEAGE` connector type**: Added `AtlanConnectorType.GENERIC_OPENLINEAGE = ("generic-openlineage", AtlanConnectionCategory.ELT)` so the SDK can target the marketplace `generic-openlineage` (GOLC) ingestion endpoint at `/events/openlineage/generic-openlineage/api/v1/lineage`. Mirrored in `pyatlan_v9`.
- **`OpenLineageEvent.emit()` accepts `connector_type`**: `emit()` and `emit_async()` now take an optional `connector_type` kwarg (defaulting to `SPARK`), matching the pattern already used by `emit_raw()` / `emit_raw_async()`. Removes the previous hardcoded SPARK assumption so callers can route events to any OpenLineage-compatible connector.

### Bug Fixes

- **`DbtMeasure` extends `SemanticMeasure`, not `Dbt`**: The Atlas type hierarchy is `DbtMeasure → SemanticMeasure → Semantic`, but the generator picked `Dbt` as the Python parent because the typedef lists `superTypes=["Dbt", "SemanticMeasure"]` (Dbt first). This broke `atlan-publish-app`'s ordering graph — `DbtSemanticModel → DbtMeasure` edges were never created via the `semantic_measures` relationship, causing publish to fail with ATLAS-404 on `semanticModel` lookup. Sibling types (`DbtDimension(SemanticDimension)`, `DbtEntity(SemanticEntity)`) already follow this pattern. Fix changes the parent class in `dbt_measure.py` and adds `_SUPERCLASS_OVERRIDES` to `class_generator.py` + a Jinja template change so future regenerations preserve the fix. Includes 7 regression tests.

### Experimental: `pyatlan_v9`

- Generated latest typedef models.

## 9.6.0 (April 27, 2026)

### New Features

- **Deprecated Phosphor icon enum values**: Added 11 deprecated Phosphor icon names (`PhActivity`, `PhCircleWavy*`, `PhFileDotted`, `PhFileSearch`, `PhFolderDotted`, `PhFolderSimpleDotted`, `PhPedestrian`, `PhTextBolder`) to `AtlanIcon` so the SDK can deserialize tag definitions that reference old icon names without failing.

### Bug Fixes

- **httpcore connection pool deadlock fix**: Fixed a production deadlock where concurrent `replace_custom_metadata` calls could hang indefinitely. Root cause: httpcore's connection pool filled with CLOSE_WAIT zombie sockets (nginx closed idle connections at 75s, httpcore didn't detect this) and `pool=None` timeout meant threads waited forever. Fix adds `pool=30.0` timeout, `keepalive_expiry=30.0` (< nginx 75s), and `response.close()` before retry sleep to release connection slots. Includes `reset_http_session()` for recovering degraded pools in long-running workflows.

### QOL Improvements

- **Connection pool config extracted to constants**: `_DEFAULT_POOL_LIMITS` and `_DEFAULT_POOL_TIMEOUT_SECONDS` defined once at module level instead of repeated 4 times across init, reset, max_retries, and timeout.
- **Trivy CI fix**: Bumped Trivy from v0.69.0 (broken download) to v0.70.0.

## 9.5.0 (April 21, 2026)

### New Features

- **Data contract delete methods**: Added `client.contracts.delete(guid)` to hard-delete all versions of a data contract and `client.contracts.delete_latest_version(guid)` to delete only the latest version and promote the previous one. Includes `extra_headers` support on `_call_api()` for per-request custom headers. Implemented across all 4 client variants (sync/async × v1/v9). See [docs](https://developer.atlan.com/snippets/datacontract/manage-via-sdks/).

### Bug Fixes

- **`showAsFeatured` silent data loss on typedef update**: `AttributeDef.Options` was missing the `show_as_featured` field, causing Pydantic to silently discard it on parse. Any SDK operation that fetched and re-submitted a Custom Metadata typedef with `showAsFeatured` set was clearing that setting in the backend. Added `show_as_featured: Optional[bool]` so the field now round-trips correctly.

### Experimental: `pyatlan_v9`

- **Data contract delete methods**: Same `delete` and `delete_latest_version` methods added to the v9 async/sync clients.
- **`show_as_featured` field**: Added to v9 `AttributeDef.Options` as well.

## 9.4.1 (April 20, 2026)

### New Features

- **Agentic type hierarchy**: Added Python SDK support for the new Agentic hierarchy (`Agentic`, `Artifact`, `Skill`, `SkillArtifact`, `Context`, `ContextRepository`, `ContextArtifact`) with NanoID-based qualifiedName patterns and auto-populated parent refs on creators. Includes `ContextLifecycleStatus` enum and `FileType.YAML`/`FileType.SQL` additions.
- **`FileType.MD` enum value**: Added `md` file type to the `FileType` enum.

### QOL Improvements

- **Generator `--override` flag**: Added `--override` flag to the generator script to force re-download of typedefs, bypassing the daily cache check.
- **Generator template coverage**: Added Jinja2 asset, attribute, and import templates for all Agentic types so the code generator can reproduce creator methods.
- **`_CORE_ASSETS` updates**: Added Agentic types, Dremio, Cloud, Google, and GCPDataplex to the core assets list to stabilize code generation layout.
- **Regenerated models from latest typedefs.**

## 9.4.0 (April 10, 2026)

### Experimental: `pyatlan_v9`

- **`BusinessPolicyException`, `BusinessPolicyIncident`, `BusinessPolicyLog` standalone classes**: Added full standalone flat asset classes for these three business policy types. Previously only `Related*` reference types existed, so users couldn't create, search, or deprecate these asset types via `pyatlan_v9`. Includes updated `__init__` exports and 12 unit tests covering imports, `type_name`, attributes, and init exports.

### QOL Improvements

- Generated latest typedef models.

## 9.3.2 (March 25, 2026)

### Bug Fixes

- **Policy duplicate detection — allow updates to proceed**: Revised the duplicate-detection logic (introduced in 9.2.0) so that legitimate policy *updates* are no longer blocked. The duplication check now only intercepts retried creations, not update requests.
- **`AtlanTag` `displayName` field**: The `AtlanTag` classification model now accepts a `displayName` field that was previously rejected during deserialization.

### Experimental: `pyatlan_v9`

- **Sync `get`/`set_custom_metadata` on `Asset`**: Added synchronous `get_custom_metadata()` and `set_custom_metadata()` methods to the v9 `Asset` class, and fixed related custom metadata integration tests.

## 9.3.1 (March 23, 2026)

### Experimental: `pyatlan_v9`

- **`typedef` typing corrections and serialization fixes**: Corrected field types in `RelationshipDef` and removed hardcoded defaults (PR #867). Optional typedef fields now use `msgspec.UNSET` so they are omitted from serialization rather than emitted as `null` (PR #870). `AttributeDef.default_value` and `index_type_es_fields` relaxed to `Any` to accommodate varied server responses (PR #872). `is_rich_text` and `custom_metadata_version` now default to `UNSET` via `__post_init__` and the `category` field is no longer silently dropped by `omit_defaults` (PR #870).

### QOL Improvements

- **Trivy action updated to `0.35.0`**: Keeps the security scanning workflow on a stable release (PR #871).

## 9.3.0 (March 16, 2026)

### New Features

- **`ModeChart.to_atlas_dict()` and `mode_report` field**: Added `to_atlas_dict()` method and `mode_report` attribute to the `ModeChart` model.

### Bug Fixes

- **`DQCondition` value type ordering**: Reordered the `Union` type for `DQCondition.value` to prioritize `int` over `str`, preventing integer values from being incorrectly coerced to strings during deserialization.
- **`RICH_TEXT` / `STRING` enum aliasing fix**: `AtlanCustomAttributePrimitiveType.RICH_TEXT` now has a distinct enum value (`"rich_text"`) instead of sharing `"string"` with `STRING`. Previously Python's Enum treated them as aliases, causing `STRING` attributes to incorrectly receive `is_rich_text=True` and RICH_TEXT multi-value restrictions to bleed into STRING attribute creation. The API-level primitive type sent in requests remains `"string"` for both (fixes PART-548).

### Experimental: `pyatlan_v9`

- **`RICH_TEXT` / `STRING` enum aliasing fix**: Same fix applied to `pyatlan_v9/model/typedef.py` — `RICH_TEXT` and `STRING` custom metadata attributes are now correctly distinguished in the v9 code path.

### QOL Improvements

- **Security guidelines for agents**: Added security guidelines documentation for agents interacting with the atlan-python SDK.

## 9.2.2 (March 11, 2026)

### QOL Improvements

- **`PackageHeaders` model and backward-compatible header emission**: Introduced a typed `PackageHeaders` model to manage package-related HTTP headers. `set_package_headers` now accepts explicit header values, and the SDK emits both `package-name` and `app-name` headers to maintain backward compatibility with older backends.
- **`/release` skill**: Added a Claude Code skill to automate SDK releases — bumps version, drafts release notes, commits, and tags.

## 9.2.1 (March 10, 2026)

### Experimental: `pyatlan_v9`

- **Restored `validate()`, `minimize()`, `relate()` SDK methods**: All generated asset types now include `validate(for_creation=False)` for dry-run field validation (checks `type_name`, `name`, `qualified_name`, optional QN pattern matching, and hierarchy-specific fields when `for_creation=True`), `minimize()` to produce a minimal updater copy, and `relate()` to create a `Related{Type}` reference. These are purely opt-in — not called by any serde path. Overlay files can override any of these methods.

## 9.2.0 (March 9, 2026)

### Breaking Changes

- **Policy duplicate detection during retries**: When a policy creation request times out but actually succeeds on the backend, the SDK now detects the existing policy instead of blindly retrying and creating a duplicate. On retry, the transport layer searches for an existing `AuthPolicy` with the same name and persona GUID — if found, it returns a mock response containing the existing policy. Code that relied on retries always creating new policies will now receive the existing policy instead.

### Experimental: `pyatlan_v9`

- **IDE autocompletion**: Replaced custom AST-based lazy loading with `lazy_loader` + Pkl-generated `.pyi` stub file. All 860+ asset classes now have full IDE type hints and autocompletion while keeping lazy import performance.
- **Pkl-generated `__init__.py` and `__init__.pyi`**: The Pkl renderer now aggregates all typedefs (via `import*` glob) to generate `__init__.py` (with `lazy_loader.attach()` mapping) and `__init__.pyi` (with explicit re-exports for IDEs) directly — no post-sync Python script needed. Hand-written types from `_init_manual.py` are included by reading the file at generation time.
- **Model updates**: Regenerated all v9 models — GTC anchor regularization, entity lineage fields (`depth`, `immediate_upstream`, `immediate_downstream`), QuickSight `useLocalTypeAsPrefix`, fully Pkl-generated `DataQualityRule`, and referenceable field descriptors via overlays.
- **`type_name` serialization fix**: `type_name` field default changed to `UNSET` so `omit_defaults=True` never omits `typeName` from API requests.
- **Simplified `transform.py`**: Removed `_normalize_camel_key()` and related camelCase abbreviation handling — msgspec structs use explicit `field(name=...)` mappings. Asset registration now uses `cls.__name__` directly.

### QOL Improvements

- Regenerated latest typedef models.
- Improved Claude Code review skill and added auto-review trigger on PR open.

## 9.1.0 (March 6, 2026)

### Breaking Changes

- `DynamoDBTable.creator()` now raises `NotImplementedError`. DynamoDB tables have a fundamentally different creation model and cannot reuse the inherited `Table.creator()`. Submit an enhancement request if you need this implemented.
- `FileClient.upload_file()` now blocks uploads of sensitive files (system paths, credential directories, hidden config files) to prevent accidental secret exposure. Users can extend the blocklist via the `PYATLAN_UPLOAD_FILE_BLOCKED_PATHS` environment variable (comma-separated substrings matched against the resolved path):
  ```
  PYATLAN_UPLOAD_FILE_BLOCKED_PATHS="/custom/secrets/,.vault,.credentials"
  ```

### QOL Improvements

- Regenerated latest typedef models.
- Upgraded Python dependencies (`pydantic`, `ruff`, `mypy`, `pytz`, and others) and GitHub Actions to latest versions.
- Replaced Sphinx with MkDocs + mkdocstrings for SDK documentation.
- Fixed circular import chains in `core/__init__.py` generator (`SQL → DbtTest`, `atlan_app` priority ordering).
- Moved `AtlanApp`, `AtlanAppTool`, `AtlanAppWorkflow`, and `AppWorkflowRun` into `core/`.
- Added Claude Code skill (`/upgrade-deps`) for automated SDK dependency maintenance.

## 8.5.3 (February 17, 2026)

### New Features

- Added `DynamoDBAttribute.creator()` methods.

### Breaking Changes

- `SSOClient.update_group_mapping()` now requires `group_map_name` as a mandatory parameter.
- Atlan tag task response no longer returns `classification_id`.

### Experimental: `pyatlan_v9` — Next-generation Pyatlan SDK powered by msgspec

Introduced `pyatlan_v9`, a ground-up rebuild of the SDK's data layer replacing Pydantic v1 with [msgspec](https://jcristharif.com/msgspec/). Ships with full functional parity — models, clients, caches, and serde — while delivering **35-50x faster** serialization performance. No implicit validators, fully typed, and extensible by design.

**Key design changes:**
- All models use `msgspec.Struct` instead of Pydantic `BaseModel`
- Explicit `UNSET` sentinel replaces Pydantic's implicit unset tracking
- `omit_defaults=True` and `rename="camel"` handle JSON wire format natively
- Zero runtime validation overhead — type safety enforced at decode time

**Benchmark results** (Python 3.11, msgspec 0.20.0, macOS):

| Operation | msgspec (v9) | Pydantic v1 | Speedup |
|---|---|---|---|
| Object creation | 1.6–2.3 us | 75–102 us | **~40-50x** |
| Serialization | 0.37–0.42 us | 14–17 us | **~35-46x** |
| Deserialization | 1.7–1.8 us | 59–65 us | **~35-36x** |
| Round-trip (ser + deser) | 2.1–2.2 us | 79–85 us | **~37-41x** |

<details>
<summary>Wall-clock at 1M operations (Connection model)</summary>

| Operation | msgspec | Pydantic v1 |
|---|---|---|
| Create 1M objects | 1.9s | 75.2s |
| Serialize 1M | 0.4s | 17.1s |
| Deserialize 1M | 1.7s | 60.6s |
| Round-trip 1M | 2.2s | 81.8s |

</details>

> `pyatlan_v9` is experimental and ships alongside the stable `pyatlan` package. APIs may change in future releases.

### QOL Improvements

- Regenerated latest SDK models.
- Switched Claude Code review model to Opus.
- Fixed flaky integration test failures across PR runs.
- Added Claude Code skill (`/generate-v9-models`) to generate and sync msgspec models for `pyatlan_v9`.


## 8.5.3 (February 17, 2026)

### QOL Improvements

- Added scheduled Trivy scan with Linear ticket creation.
- Added Claude Code workflow with /review skill for PR reviews.

## 8.5.2 (February 13, 2026)

### QOL Improvements

- Added Harbor publish along with existing GHCR image publish.
- Added requirements.txt and pre-commit sync for Snyk dependency scans.
- Added Trivy image + uv.lock scans for PRs and main.
- Updated Docker docs to reference Harbor pulls.
- Added local Snyk scan instructions.
- Limited uv resolution to CPython and bump urllib3 to fix CVEs.

## 8.5.1 (January 29, 2026)

### Bug Fixes

- Fixed iterative pagination in User and Role caches that prevented all records from being properly populated during cache refresh operations.

### QOL Improvements

- Enhanced SDK client retry mechanism to automatically retry HTTP `302` redirect responses, improving reliability when handling temporary redirects.
- Updated admin integration tests to align with latest backend API changes.


## 8.5.0 (January 14, 2026)

### New Features

- Added support for new DQ rule templates.

### Breaking Changes

- `DataQualityRule.table_level_rule_creator()` and `DataQualityRule.column_level_rule_creator()` now accept `rule_type` as a `DataQualityRuleTemplateType` enum instead of `str`.
- `threshold_compare_operator` is now an optional parameter for `DataQualityRule.table_level_rule_creator()`.

## 8.4.6 (January 13, 2026)

### New Features

- Added support for OAuth client management (`create`, `get`, `get_by_id`, `update`, `purge`).
- Added optional `field_type` parameter to `SearchableField.has_any_value()`, allowing users to handle cases when text exceeds **5K characters** (the keyword field on an attribute can be empty while the text field on the same attribute is populated).

### QOL Improvements

- Regenerated models with latest typedef definitions.
- Published `pyatlan` with Chainguard golden image for each new release.

## 8.4.5 (December 15, 2025)

### QOL Improvements

- Regenerated models with latest typedefs.

## 8.4.4 (December 3, 2025)

### New Features

- Added support for initializing `AtlanClient` and `AsyncAtlanClient` using `OAuth` credentials.

### Bug Fixes

- Fixed `pyatlan.test_utils.create_connection()` which was missing the required `client` parameter.

## 8.4.3 (December 2, 2025)

### Bug Fixes

- Fixed `KeywordField` for `AtlanTask.GUID`.
- Fixed circular imports caused by Process-SQL asset relationships in generated typedef models.

### QOL Improvements

- Regenerated models with latest typedefs.

## 8.4.2 (November 13, 2025)

### New Features

- Added DQ permissions enum.

### Bug Fixes

- Fixed time range condition in `WorkflowClient.find_runs_by_status_and_time_range()`

## 8.4.1 (November 10, 2025)

### New Features

- Added support for `enable_full_restriction` to `IndexSearchRequest` & `FluentSearch`.
- Added support for new custom metadata attribute type: `AtlanCustomAttributePrimitiveType.RichText`.

### Bug Fixes

- Fixed field type for `LineageListRequest.relation_attributes` to be `Optional[List[str]]`.

### QOL Improvements

- Updated source tag search with value query to match latest frontend changes.

## 8.4.0 (October 30, 2025)

### Bug Fixes

- Fixed source tag search with value.
- Fixed missing `needs_type_filter ` check on asset  `superTypeName`.

### QOL Improvements

- Bumped SDK core and development dependencies to the latest version.

## 8.3.1 (October 16, 2025)

### Bug Fixes

- Fixed permission errors (`403`) by using the `/whoami` endpoint instead of Keycloak endpoints (which are blocked for customers) to fetch user role mapping for determining whether an API token is being used.

## 8.3.0 (October 15, 2025)

### Bug Fixes

- Fixed an issue with populating data to raw arguments in data quality.

### Experimental

- Added new `httpx` transports: `PyatlanSyncTransport` and `PyatlanAsyncTransport` that support both `httpx` retries and gracefully handle proxy configuration for the SDK client. Previously, we found that `RetryTransport` (from `httpx-retries`) clobbered proxy environment variables, causing users to face SSL certificate verification failures and EOF errors.

Users can explicitly configure these settings via:
```py
from pyatlan.client.atlan import AtlanClient

proxy_settings = {
    "verify": "mitmproxy-ca-cert.pem",  # Path to certificate file
    "proxy": "http://127.0.0.1:8081",    # Proxy URL to use
}

client = AtlanClient(**proxy_settings)
```

Or through environment variables:
```bash
export HTTP_PROXY=http://127.0.0.1:8080
export HTTPS_PROXY=http://127.0.0.1:8080
export SSL_CERT_FILE=mitmproxy-ca-cert.pem
```

## 8.2.2 (October 7, 2025)

### Features

- Added support for new workflow package: `WorkflowPackage.CLOUDSQL_POSTGRES`.
- Added support for new connector type: `AtlanConnectorType.CLOUDSQL_POSTGRES`.
- Added support for executing following workflow methods using AtlanClient created via OAuth credentials.

1. WorkflowClient.rerun()
2. WorkflowClient.run()
3. WorkflowClient.update()
4. WorkflowClient.delete()
5. WorkflowClient.add_schedule()
6. WorkflowClient.remove_schedule()

### QOL Improvements

- Fixed snyk reported code issues.

## 8.2.1 (September 30, 2025)

### Bug Fixes

- Fixed issue where custom metadata was not flushed before `BulkRequest` object creation for the `save()` operation in `AssetClient.save_merging_cm()` method.

## 8.2.0 (September 25, 2025)

### Breaking Changes

- DQ typedef models have been updated, resulting in changes to the DQ rule management methods. Please refer to the developer docs for the latest updates: https://developer.atlan.com/patterns/create/dq_rules

### QOL Improvements

- Generated latest DQ typedef models.
- Updated `Meaning` model fields to be optional.
- Forced lowercase when creating custom connector names to avoid frontend-related issues.

## 8.1.1 (September 23, 2025)

### New Features

- Added support for `Databricks` system tables in the `DatabricksCrawler`.
- Added support for rule conditions and incremental updates for `DQ` rules.

### Other Changes

- Deprecated `DatabricksCrawler.basic_auth()` - use `pat()` authentication method instead.

## 8.1.0 (September 15, 2025)

### New Features

- Added support for sending OpenLineage raw events via `OpenLineageEvent.emit_raw()`

### Breaking Changes

- Relationship removal is now idempotent, i.e removing a relationship that does not already exist using `SaveSemantic.REMOVE` will now simply succeed and do nothing, whereas it previously threw a `NotFoundError`

### QOL Improvements

- Added update policy test to `persona_test`
- Fixed `glossary_test.test_remove_unrelated_relationship`
- Updated `Atlas` endpoint to enforce rate limiting within apps

## 8.0.2 (September 1, 2025)

### Bug Fixes

- Fixed `AsyncWorkflowSearchResponse` async iterator.

## 8.0.1 (August 25, 2025)

### New Features

- Added `get_client_async()` for initializing `AsyncAtlanClient` when using packages.
- Added optional parameter `set_pkg_headers()` to set package headers on the client (defaults to `False`).

### Bug Fixes

- Fixed `httpcore.LocalProtocolError` exception in `get_client()` that occurred when `ATLAN_API_KEY` environment variable was not configured (commonly encountered during package impersonation with empty `API` key strings).

### QOL Improvements

- Added `pyatlan-wolfi-base` Docker image workflow with enhanced capabilities:
  - **Automatic release builds**: Workflow now triggers automatically on GitHub releases using latest Python (3.13) and SDK versions
  - **Git branch support**: Added ability to install SDK from any git branch for development/testing purposes via `pyatlan_branch` parameter
  - **Smart installation logic**: Automatically chooses between PyPI (stable) or git (development) installation methods
  - **Enhanced tagging**: Branch builds tagged as `branch-{branchname}-{python}-{commit}` for easy identification
  - **Build metadata**: Images include labels tracking installation source, version/branch, and Python version
  - **Conditional PyPI checks**: Skips PyPI availability checks when installing from git branches
  - **Improved logging**: Shows installation method, branch info, and trigger source in build outputs

## 8.0.0 (August 20, 2025)

### New Features

#### Full `async/await` support

- Added `async/await` support to the SDK. All existing `clients` and `caches` now have async variants, plus new models (mainly search result models that require `async` iteration). Following `aio` directory convention for all async components.
- Implemented `AsyncAtlanClient` for all async operations (extends `AtlanClient` for reusability).
- For methods that accept client parameters, we've added corresponding `*_async()` variants:

| sync method | async method |
|------------|--------------|
| `Connection.creator()` | `Connection.creator_async()` |
| `Badge.creator()` | `Badge.creator_async()` |
| `FluentSearch.execute()` | `FluentSearch.execute_async()` |
| `AtlanTag.of()` | `AtlanTag.of_async()` |
| `SourceTagAttachment.by_name()` | `SourceTagAttachment.by_name_async()` |
| `CompoundQuery.tagged_with_value()` | `CompoundQuery.tagged_with_value_async()` |
| `Referenceable.json()` | `Referenceable.json_async()` |
| `Referenceable.get_custom_metadata()` | `Referenceable.get_custom_metadata_async()` |
| `Referenceable.set_custom_metadata()` | `Referenceable.set_custom_metadata_async()` |
| `Referenceable.flush_custom_metadata()` | `Referenceable.flush_custom_metadata_async()` |

#### Shared business logic architecture

- Extracted common functionality (request preparation and response processing) into a separate `common` sub-package. This enables reuse across both sync and async operations - only the middle layer (API calling with respective clients) differs.

Example:
```python
from pyatlan.client.common import FindPurposesByName

@validate_arguments
async def find_purposes_by_name(
    self,
    name: str,
    attributes: Optional[List[str]] = None,
) -> List[Purpose]:
    """
    Async find purposes by name using shared business logic.
    :param name: of the purpose
    :param attributes: (optional) collection of attributes to retrieve for the purpose
    :returns: all purposes with that name, if found
    :raises NotFoundError: if no purpose with the provided name exists
    """
    search_request = FindPurposesByName.prepare_request(name, attributes)
    search_results = await self.search(search_request)
    return FindPurposesByName.process_response(
        search_results, name, allow_multiple=True
    )
```

### Documentation

- **Asynchronous SDK operations**: https://developer.atlan.com/sdks/python/#asynchronous-sdk-operations

### Breaking Changes

While these aren't direct breaking changes to the SDK API, they may affect your code if you depend on these libraries:

- **Migrated from [`requests`](https://requests.readthedocs.io/en/latest) to [`httpx`](https://www.python-httpx.org)**: Completely removed support for `requests` library and migrated to `httpx`, which provides similar API for `sync` operations plus `async` client support for async operations.
- **Replaced [`urllib3`](https://urllib3.readthedocs.io/en/stable) with [`httpx-retries`](https://will-ockmore.github.io/httpx-retries)**: Removed support for `urllib3` retry mechanism and implemented retries using `httpx-retries` library (API remains similar).

### QOL Improvements

- Generated latest typedef models.
- Removed redundant requirements files (no longer needed since migration to `uv` in previous releases).
- Updated GitHub workflows for Docker image builds to use `uv sync` (without dev dependencies).
- Added unit and integration tests for `async` SDK.
- Added `x-atlan-client-type: sync` or `x-atlan-client-type: async` to SDK headers and logging for better observability.
- Added `async-integration-tests` job to `pyatlan-pr.yaml` workflow. Triggers when there are changes to async SDK code or can be triggered manually via `run-async-tests` label on PR.
- Async SDK unit tests run by default on every commit push as they have minimal time impact on test suite.
- Used module-scoped asyncio test fixtures similar to sync integration tests:
  ```toml
  # pyproject.toml
  asyncio_mode = "auto"
  asyncio_default_test_loop_scope = "module"
  asyncio_default_fixture_loop_scope = "module"
  ```
- Used `token_client` fixtures when creating/deleting API tokens with `retry=0` to avoid token overpopulation in test tenants.

## 7.2.0 (August 13, 2025)

### New Features

- Added Python version info to SDK headers and API logger.

### Breaking Changes

- Dropped support for `Python 3.8`.

### QOL Improvements

- Bumped various `core` and `dev` dependencies to latest versions.
- Fixed conda publish and pyatlan publish GitHub workflows.
- Fixed integration test failures due to Elasticsearch eventual consistency using custom retries (`glossary_test`, `test_client`, `persona/purpose_test`)

## 7.1.6 (August 7, 2025)

### Experimental

- Added support for adding DQ rules schedule to an asset (i.e: `AssetClient.add_dq_rule_schedule()`).

## 7.1.5 (August 1, 2025)

### Experimental
- Added `creator()` and `updater()` methods for `alpha_DQRule` (data quality rule) asset.

### QOL Improvements
- Generated latest typedef models.

## 7.1.4 (July 29, 2025)

### New Features
- Added new connector types: `AI`, `SAP_ECC`.
- Added new connector categories: `AI`, `ERP`.
- Added `creator()` methods for AI-based assets.
- Added support for `applicable_ai_asset_types` to custom metadata `AttributeDef.Options`.

### Experimental
- Added support for `AtlanClient` initialization via API token `guid`.

### Bug Fixes
- Fixed handling of `source_tag` when no `attributes` present.

### QOL Improvements
- Generated latest typedef models.
- Updated Dockerfile to use `python:3.13.5-slim-bookworm`.
- Removed unused `requirements` files (now using `pyproject.toml`).

## 7.1.3 (July 21, 2025)

### Bug Fixes

- Fixed multi-value select bug in the CM attribute setter.

## 7.1.2 (July 11, 2025)

### QOL Improvements

- Fixed `suggestions_test` integration test.
- Generated latest typedefs models.
- Updated Dockerfile to use `python:3.11-slim`.
- Migrated SDK to use [`uv`](https://docs.astral.sh/uv) for development, building, and publishing the SDK.
- Added `delete_type:HARD` optional parameter to `AssetClient.purge_by_guid()` method.
- Made `FluentSearch` usage in `get_by_*()` methods consistent with `GET` API (return `active` and `archived` assets).

## 7.1.1 (June 30, 2025)

### QOL Improvements

- Regenerated latest typedef models.

## 7.1.0 (June 27, 2025)

###  New Features

- Added support for managing asset-to-asset relationships that include attributes. Atlan now allows setting attributes on specific relationship types, providing richer metadata and contextual information for how assets are connected.

- These relationships can now carry attributes, similar to how assets themselves have attributes. The SDK has been enhanced to deserialize, access, and serialize these relationship-level attributes correctly.

**Supported relationship types:**

1. `AtlasGlossaryAntonym`
2. `AtlasGlossarySynonym`
3. `AtlasGlossaryReplacementTerm`
4. `AtlasGlossarySemanticAssignment`
5. `AtlasGlossaryPreferredTerm`
6. `AtlasGlossaryRelatedTerm`
7. `AtlasGlossaryTermCategorization`
8. `AtlasGlossaryTranslation`
9. `AtlasGlossaryValidValue`
10. `AtlasGlossaryIsARelationship`
11. `CustomParentEntityCustomChildEntities`
12. `CustomRelatedFromEntitiesCustomRelatedToEntities`
13. `UserDefRelationship`


### Breaking Changes

`Referenceable.relationship_attributes` field type updated:

This field now supports:

  ```python
  Optional[Union[RelationshipAttributes, Dict[str, Any]]]
  ```

- If `typeName` is present in the relationship attributes, the SDK will attempt to deserialize using the correct subclass of `RelationshipAttributes`.
- If `typeName` is absent, it will fall back to a raw `Dict[str, Any]`.

  **Before:**

  ```python
  Optional[Dict[str, Any]]
  ```

- This change improves structure and typing for known relationship models while maintaining backward compatibility via fallback.

### Bug Fixes

- Fixed automatic `401` token refresh by using `ContextVar` for `AtlanClient._401_has_retried` (bool flag) to avoid race conditions when executing SDK code in multithreading environments.

## 7.0.1 (June 16, 2025)

### Bug Fixes

- Fixed `FluentSearch` criteria used for `get_by_qualified_name/guid` method to include asset `type_name`.

## 7.0.0 (June 12, 2025)

### New Features

- Added `long` as a new primitive type in the `AtlanCustomAttributePrimitiveType` enum.
- Support for `Unmodeled` Asset types: Introduced `IndistinctAsset` as a fallback model to handle asset types that are not explicitly modeled in the SDK. Previously, such assets would return `None` (e.g: for newly introduced typedefs). With this release, the SDK will now return an `IndistinctAsset` instance containing basic fields such as `guid`, `qualifiedName`, and `typeName`.

### Bug Fixes

- Fixed timestamp-based pagination in search log results when `from + size` exceeds the Elasticsearch window size (10,000 records) due to identical creation times for the first and last records in a retrieved page.

### Breaking Changes

This release includes a **major refactor** that eliminates all usage of `AtlanClient.get_current_client()` and `set_current_client()` (previously implemented using [ContextVar](https://docs.python.org/3/library/contextvars.html) / [thread-local storage](https://docs.python.org/3/library/threading.html#thread-local-data) a.k.a TLS). Although the earlier design was thread-safe, it still required users to explicitly use `PyAtlanThreadPoolExecutor` in multi-threaded environments, resulting in unintuitive patterns and frequent client initialization errors.

The new design simplifies usage by requiring the `client` to be passed explicitly to SDK operations that interact with the Atlan platform.

> [!IMPORTANT]
> ### Affected Areas
> All the following methods or constructors now require a `client` argument due to the removal of `AtlanClient.get_current_client()`:

##### `pyatlan.model.assets`:

- `Referenceable.flush_custom_metadata(client)`
- `Referenceable.get_custom_metadata(client, name)`
- `Referenceable.set_custom_metadata(client, custom_metadata)`
- `CustomMetadataProxy` is no longer initialized in `Referenceable.__init__()` (as it now requires a `client`).
- `Purpose.create_metadata_policy()`
- `Purpose.create_data_policy()`
- `DataProduct.get_assets()`
- `Badge.creator()`
- `Connection.creator()`

> [!NOTE]
> `user_cache.validate_names`, `role_cache.validate_idstrs`, and `group_cache.validate_aliases` are now invoked inside `Connection.creator()` instead of field validators.

##### `pyatlan.model.custom_metadata`:

- `CustomMetadataDict(client, name)`
- `CustomMetadataProxy(client, business_attributes)`

##### `pyatlan.model.structs`:

- `SourceTagAttachment.by_name()`
- `SourceTagAttachment.by_qualified_name()`
- `SourceTagName()` (constructor)

##### `pyatlan.model.suggestions`:

- `Suggestions.get()`
- `Suggestions.apply()`

##### `pyatlan.model.fluent_search`

- `FluentSearch.tagged()`
- `FluentSearch.tagged_with_value()`

##### `pyatlan.client.typedef`

- `EnumDef.update()`
- `AttributeDef.create()`

##### `pyatlan.model.search`

- `Exists.with_custom_metadata()`
- `Term.with_custom_metadata()`

##### `pyatlan.client.atlan`

- `client_connection()`

##### `pyatlan.model.open_lineage`

- `OpenLineageEvent.emit()`

##### `pyatlan.model.fields.atlan_fields`

- `CustomMetadataField()` (constructor)

##### `pyatlan.model.packages`

- All crawler class constructors

### Serialization & Deserialization

#### API Serialization:

- Now handled via the `AtlanRequest` wrapper, which takes a Pydantic model instance and a client.
- It automatically performs translation (e.g: converting human-readable Atlan tag names into hashed IDs using `AtlanTagRetranslator`).

#### API Response Deserialization:

- Responses are processed using the `AtlanResponse` wrapper.
- It translates raw JSON into readable formats via registered translators like `AtlanTagTranslator` (e.g: converting hashed tag IDs into human-readable names).

### AtlanTag / AtlanTagName Changes

- **Human-readable tag names** are now only available when deserializing through `AtlanResponse` (which requires a valid `client`). If skipped, tag names remain in **hashed ID format** or as plain strings.
- Deleted tags (e.g `AtlanTagName('(DELETED)')`) will only appear when using `AtlanResponse`, as the lookup requires a client to determine tag validity.
- The `.id` attribute on `AtlanTagName` has been **deprecated**. Use `AtlanTag.tag_id` to access the hashed ID instead.
- Fixed typo in `AtlanTag.source_tag_attachments` field name.

### Other Changes

- `AtlanClient._401_has_retried` is now marked as a `PrivateAttr`.
- `IndexSearchRequest.Metadata` has been moved to a separate class: `IndexSearchRequestMetadata`.
- Removed deprecated methods:

  - `AtlanClient.get_current_client()`
  - `AtlanClient.set_current_client()`
  - `PyAtlanThreadPoolExecutor` (SDK is now fully thread-safe without it)
- Updated `AssetClient.find_domain_by_name()` and `AssetClient.find_product_by_name()` to return only active `Domain` and `Product` assets. Previously, these methods returned both `active` and `archived` assets, which caused issues when assets with the same `name` existed in both states - the first match (possibly `archived`) was returned.

### QOL Improvements

- Regenerated latest typedef models.
- Refactored integration and unit tests to eliminate reliance on `AtlanClient.get_current_client()` / `set_current_client()`.

## 6.2.1 (May 30, 2025)

### Bug Fixes

- Extended the `applicable_asset_types` attribute definition to accept `Union[Set[str], AssetTypes]`.
- Fixed validation to ensure provided asset type names match existing SDK asset classes, replacing the previously hard-coded set of asset types.

### QOL Improvements

- Regenerated the latest typedef models.

## 6.2.0 (May 27, 2025)

### New Features

- Added support for `append_atlan_tags` in the `AssetClient.save()` method.
- Introduced `AssetClient.remove_atlan_tags()` to allow users to remove one or more Atlan tags from a given asset.
- Enhanced the SDK to use the `/bulk` endpoint for all tag mutation operations.

### Breaking Changes

- All tag mutation methods (`add_atlan_tags`, `update_atlan_tags`, `remove_atlan_tag`, `remove_atlan_tags`) now use the `/bulk` endpoint. As a result, these methods will now return an `AssetMutationResponse` instead of `None`, reflecting the updated API response structure.

### Bug Fixes

- Added default `typeName=Referenceable` to `Bool`(must) clauses when `typeName` is not explicitly provided in the search request. This ensures the SDK retrieves only assets that inherit from the `Referenceable` supertype.

### QOL Improvements

- Regenerated the latest typedef models.

## 6.1.1 (May 21, 2025)

### New Features

- Added a utility method (`AssetClient.process_assets()`) to simplify processing (e.g: updating) assets while iterating through search results.

### Bug Fixes

- Fixed an issue where the search pagination loop would break due to invalid assets.
- If `typeName` is not explicitly provided in the search request, the SDK now defaults to retrieving only assets with the `Referenceable` supertype. This avoids including non-asset records in the response.

### QOL Improvements

- Regenerated the latest typedef models.
- Added a retry loop to ensure the token is fully active before retrying the original request after a `401` response.

## 6.1.0 (May 13, 2025)

### New Features

- Added support for dynamic extension of `AtlanConnectorType` to enable custom connectors.
- Migrated from `TLS` to `ContextVars` to support both multithreaded and asynchronous environments.
- Introduced `PyAtlanThreadPoolExecutor` in `pyatlan.utils`, which preserves context variables across threads—useful for running SDK methods in multithreaded or asynchronous environments.
- Implemented iterative pagination in `WorkflowClient` search methods.

### Bug Fixes

- Resolved pagination issues in `UserClient` and `GroupClient` search methods.

### QOL Improvements

- Regenerated the latest typedef models.
- Simplified the `creator()` methods in both `AtlasGlossaryTerm` and `AtlasGlossaryCategory`.

## 6.0.6 (April 29, 2025)

### New Features

- Added optional `description` parameter to `CustomMetadataDef/AttributeDef.create()` methods.

### QOL Improvements

- Added issue and pull request templates.
- Added support for [`vcrpy`](https://vcrpy.readthedocs.io/en/latest) to enable easy mocking/patching of HTTP interactions, especially useful for 3rd-party API testing.

## 6.0.5 (April 25, 2025)

### New Features

- Added a public method `AtlanClient.init_for_multithreading()` to allow users to prepare the given client for use in multi-threaded environments.

### QOL Improvements

- Regenerated the latest typedef models.

## 6.0.4 (April 22, 2025)

### New Features

- Added a new optional parameter `source_tag_qualified_name` to `tagged_with_value()`, allowing users to explicitly provide the qualified name of the source tag to match (useful when multiple tags share the same name).
- Added support for undefined fields in `DataContract` models by allowing extra fields in `AtlanYamlModel` (`Extra.allow`).
- Added a new method `WorkflowClient.find_runs_by_status_and_time_range()` to search workflow runs based on their status and a specified time range (`started_at` / `finished_at`).

### QOL Improvements

- Regenerated the latest typedef models.
- Fixed an indentation issue in the `DocumentDB` generator Jinja templates.

## 6.0.3 (April 15, 2025)

### New Features

- Added `creator()` methods for `DocumentDB` assets.
- Added support for `purpose` and `persona` in `IndexSearchRequest`.
- Added an optional `workflow_name` parameter to the `WorkflowClient.monitor()` method.
- Added an optional `workflow_name` parameter to the `CredentialClient.get_all()` method.

### QOL Improvements

- Generated the latest typedef models.

## 6.0.2 (April 07, 2025)

### QOL Improvements

- Updated SDK generator templates to align with `Pyatlan v6` changes.
- Added a delay after token refresh to avoid empty (`[]`) typedef responses.

### Bug Fixes

- Reset `has_retried` if the last retry wasn't a `401` to allow future token refresh attempts.

## 6.0.1 (March 28, 2025)

### QOL Improvements

- Migrated the boolean field `AtlanClient.401_tls.has_retried` to thread-local storage (`TLS`) to enhance `401` token handling in a multithreaded environment.

## 6.0.0 (March 20, 2025)

### New Features

- Added a new connector type: `DOCUMENTDB`.

### Breaking Changes

- `DataProduct.get_assets()` method now raises `InvalidRequestError` when there is a missing value for `data_product_assets_d_s_l`, which is required to retrieve product assets.
- Fixed SDK cache inconsistencies and unexpected behavior when running in concurrent/multi-threaded environments.

  - Completely migrated from `AtlanClient._default_client` to `AtlanClient._current_client_tls` (which uses thread-local storage) to prevent sharing this class variable across multiple threads. Previously, it was shared across threads, resulting in inconsistent behavior in SDK caches.
  - Removed `cache_key` maintenance that used to maintain cache instances per Atlan client hash (`cache_key(base_url, api_key)`).
  - Now, all caches are bound to an `AtlanClient` instance, requiring the migration of all cache methods from class methods to instance methods.
  - Caches remain tracked even in cases of automatic token refresh for the client.

  The following example illustrates the migration:

  ### Before

  ```py
  from pyatlan.cache.atlan_tag_cache import AtlanTagCache

  c1 = AtlanClient()
  tag_id = AtlanTagCache.get_id_for_name(atlan_tag_name)  # <-- Uses default client (c1), populates the caches (API call), and uses cache_key to store the cache instance
  tag_id = AtlanTagCache.get_id_for_name(atlan_tag_name)  # Returns the ID from the cache (no API call)

  c2 = AtlanClient()
  tag_id = AtlanTagCache.get_id_for_name(atlan_tag_name)  # <-- Uses default client (c2), populates the caches, and uses cache_key to store the cache instance
  tag_id = AtlanTagCache.get_id_for_name(atlan_tag_name)  # Returns the ID from the cache (no API call)

  c1 = AtlanClient()
  tag_id = AtlanTagCache.get_id_for_name(atlan_tag_name)  # <-- c1 initialized again. Since cache_key was used for c1 previously, the populated cache instance in memory is reused, avoiding an API call.
  tag_id = AtlanTagCache.get_id_for_name(atlan_tag_name)  # Returns the ID from the cache (no API call)
  ```

  ### Now (caches are bound to the client and maintained only upon the first client initialization):

  ```py
  c1 = AtlanClient()

  tag_id = c1.atlan_tag_cache.get_id_for_name(atlan_tag_name)  # <-- Uses default client (c1) and populates the caches (API call)

  # OR
  tag_id = AtlanClient.get_current_client().atlan_tag_cache.get_id_for_name(atlan_tag_name) # <-- Uses default client (c1) and populates the caches (API call)
  tag_id = AtlanTagCache.get_id_for_name(atlan_tag_name)  # Returns the ID from the cache (no API call)

  c2 = AtlanClient()
  tag_id = c2.atlan_tag_cache.get_id_for_name(atlan_tag_name)  # <-- Uses default client (c2) and populates the cache (API call)
  tag_id = c2.atlan_tag_cache.get_id_for_name(atlan_tag_name)  # Returns the ID from the cache (no API call)

  c1 = AtlanClient()
  tag_id = c1.atlan_tag_cache.get_id_for_name(atlan_tag_name)  # <-- c1 initialized again. Since no cache_key is used in the latest approach, the previously populated cache instance is gone, and we need to make an API call to populate the cache for c1.
  tag_id = c1.atlan_tag_cache.get_id_for_name(atlan_tag_name)  # Returns the ID from the cache (no API call)
  ```

## 5.0.2 (March 11, 2025)

### New Features

- Added handling for `doc` and `errorDoc` in `AtlanError`.
- Introduced secure agent support in `OracleCrawler`.

### QOL Improvements

- Fixed vulnerabilities reported by `dependabot`.

## 5.0.1 (March 03, 2025)

### New Features

- Added support for the `MatchPhase` query.

### Bug Fixes

- Fixed handling of `aggregations` in `AuditSearchResults`.
- Added `Asset.TYPE_NAME` condition in `FluentSearch` for append terms, replace terms, and remove terms methods to ensure an exact asset match. Previously, only `Asset.GUID/QUALIFIED_NAME` was used, which could return multiple assets if different asset types had the same qualified name.

### QOL Improvements

- Bumped `ruff` to the latest version (`0.9.9`).
- Removed redundant asset relationship fields.
- Updated the **"Installing for Development"** section in `README.md`.

## 5.0.0 (February 26, 2025)

### New Features

- Added a new method `DataProduct.get_assets()` that retrieves a list of all assets linked to the provided data product.
- Made `DSL.query` and `DSL.post_filter` more flexible to accept both raw dictionaries and Python objects, i.e: `Optional[Union[Dict[str, Any], Query]]`. This change allows users to construct `DSL` objects directly with raw DSL dictionaries, e.g: `DSL(**raw_dsl)`.
- Added `creator_with_prefix()` methods to `GCSObject` and `ADLSObject`.
- Added a model utility method (`construct_object_key()`) to construct a consistent `objectKey` for object store assets, which is currently used by `creator_with_prefix()`.

### Bug Fixes

- Fixed the `creator_with_prefix()` method to use an empty prefix `""` (representing the root path) for object store assets.
- Fixed `SearchLogResults` edge cases — when `entityGuidsAll` is empty `[]` or `"undefined"` in the request `DSL`.

### Breaking Changes

- `bucket_name` and `adls_container_name` are now mandatory for `GCSObject.creator()` and `ADLSObject.creator()`, respectively.
- Changed the default behavior for tag propagation to `False` by default. This means:
  - `remove_propagations_on_entity_delete` is now `True` by default.
  - `propagate`, `restrict_propagation_through_lineage`, and `restrict_propagation_through_hierarchy` are now `False` by default.

### QOL Improvements

- Generated the latest typedef models.
- Upgraded SDK dependencies (main, dev, and tooling) to the latest versions.
- Replaced the existing linter and formatter stack (`black`, `flake8`, `isort`, `autoflake8`) with `ruff`, significantly reducing development and QA check times while consolidating all configurations into a single file.

## 4.2.5 (February 19, 2025)

### New Features

- Added the SDK version number to the `_api_logger()` method.
- Added handling of the `errorCause` property to `AtlanError`.
- Added `creator()` methods for `Quicksight` assets.
- Added support for new connector types:
  - `KX`
  - `CRATEDB`

### QOL Improvements

- Updated GitHub Actions to the latest version.
- Fixed broken integration tests for adding, replacing, and removing glossary terms.
- Added a new job to the `pyatlan-publish.yml` workflow to publish the SDK image to Docker Hub.
- Introduced `dependabot.yml` to manage dependencies and actions, and removed unused secret variables.

## 4.2.4 (February 13, 2025)

### Bug Fixes

- Added safeguards for search pagination methods.

## 4.2.3 (February 13, 2025)

### QOL Improvements

- Refactored the `Asset._convert_to_real_type_()` method for improved clarity and robustness.
  In certain cases, asset types may be undefined or invalid—likely due to an unhandled backend edge case. To handle this gracefully, the method now ignores deserialization of `Asset` API responses in such scenarios instead of raising `AttributeError` or `TypeError` exceptions.

## 4.2.2 (February 11, 2025)

### New Features

- Added support for `AnaplanSystemDimension.creator()`.
- Added support for new connector types:
  - `RDS`
  - `SHARE_POINT`
  - `SHARED_DRIVE`

### QOL Improvements

- Generated the latest typedef models.
- Upgraded pre-commit `black` to `24.4.2`
- Enhanced `Credential.creator()` validation to prevent creating fake credentials when `test=False`.

## 4.2.1 (February 6, 2025)

### New Features

- Added an optional boolean parameter `test=True` to `CredentialClient.creator()` method.

## 4.2.0 (February 4, 2025)

### New Features

- Added support for `ApplicationField.creator()`.
- Added support for `CredentialClient.creator()`.
- Added support for optional parameters (`host` and `port`) in `Connection.creator()`.

### Breaking Changes

- Changed the default setting for tracking `IndexSearch` logs (`save_search_log=False`). Previously, it was set to `True`, which was causing frequent Out of Memory (OOM) issues in Metastore pods.

### QOL Improvements

- Generated the latest typedef models.
- Refactored `OpenLineageClient.create_connection()` to use the new `CredentialClient.creator()`.

## 4.1.0 (January 28, 2025)

### New Features

- Added support for the following workflow packages:

  **Connectors**
  - OracleCrawler

  **Utils**
  - LineageBuilder
  - LineageGenerator
  - APITokenConnectionAdmin

- Extended the `WorkflowClient.run()` method to accept raw workflow `JSON` strings.

### Bug Fixes

- Updated the maximum API token Time-to-Expire (TTE) to `5` years. Previously, the value was set to `13` years (`409,968,000` seconds), which was reverted to `5` years due to an integer overflow issue in Keycloak. For more details, see [Keycloak Issue #19671](https://github.com/keycloak/keycloak/issues/19671).

### QOL Improvements

- Added support for Python `3.10`, `3.11`, `3.12` and `3.13`.
- Increased the default read timeout for `AtlanClient` to `900` seconds (`15` minutes).

## 4.0.2 (January 22, 2025)

### New Features

- Added the ability to the custom package logger to log unhandled exceptions.
- Added support for creating `OpenLineage` connections (`OpenLineageClient.create_connection()`).

### Bug Fixes

- Fixed handling of deleted `AtlanTagName`s in `Purpose` asset deserialization. Previously, a `ValueError` was raised.

### QOL Improvements

- Added an "Installing for development" section to the `README.md`.
- Removed `pyatlan-codeql.yaml` in favor of using the organization-level CodeQL workflow to avoid configuration conflicts and failures.

## 4.0.1 (January 14, 2025)

### New Features

- Added a new connector type (`CUSTOM`).
- Added support for the `DatabricksCrawler` and `DatabricksMiner` workflow packages.
- Added the `creator()` method for the following assets:
  - `Custom`
  - `Dataverse`
  - `TablePartition`

### Bug Fixes

- Fixed `_user_id` handling issue in `pyatlan.pkg.utils.get_client()`.

### QOL Improvements

- Enhanced test coverage with additional unit tests for the `append_terms`, `replace_terms`, and `remove_terms` methods.

## 4.0.0 (January 7, 2025)

### New Features

- Added support for the `MongoDBCrawler` workflow package.
- Added `creator()` and `updater()` methods for the `Procedure` asset.

### Bug Fixes

- Fixed `WorkflowClient.find_by_type()` method to use regular expressions.
- Fixed `AttributeDef.cardinality` and `AttributeDef.type_name` handling for the new `multi_value_select` attribute option.
- Fixed issues with the `AssetClient.append_terms()`, `remove_terms()`, and `replace_terms()` methods:
  - These methods now use `SaveSemantic.APPEND/REPLACE/REMOVE`, which is more optimized and faster than the previous approach that required full asset retrieval with relationships for updates.
- Fixed `S3Object.create_with_prefix()` and `creator()` to ensure the `s3_bucket_name` field is correctly set.

### Breaking Changes

- The default behavior of `AssetClient.get_by_guid()` and `AssetClient.get_by_qualified_name()` has changed:

  - By default, these methods now ignore relationships during asset retrieval (`ignore_relationships: bool = True`). Previously, this was set to `False`.
  - This change is intentional, aiming to prevent users from retrieving more information than necessary, thereby reducing the overall runtime of requests.
  - Users can now use the `attributes` and `related_attributes` optional parameters to specify the exact details required for each search result. This ensures precise and efficient searches. See:
  [Advanced Examples - Read](https://developer.atlan.com/snippets/advanced-examples/read).

  ```python
  # In this example, we are retrieving the "userDescription" attribute
  # for both the glossary and its terms (related attribute).
  # You can also retrieve other attributes as illustrated below:

  glossary = client.asset.get_by_guid(
        guid="b4113341-251b-4adc-81fb-2420501c30e6",
        asset_type=AtlasGlossary,
        min_ext_info=False,
        ignore_relationships=True,
        attributes=[AtlasGlossary.USER_DESCRIPTION, AtlasGlossary.TERMS],
        related_attributes=[AtlasGlossaryTerm.USER_DESCRIPTION]
    )

  glossary = client.asset.get_by_qualified_name(
      asset_type=AtlasGlossary,
      qualified_name="pXkf3RUvsIOIG8xnn0W3O",
      min_ext_info=False,
      ignore_relationships=True,
      attributes=[AtlasGlossary.USER_DESCRIPTION, AtlasGlossary.TERMS],
      related_attributes=[AtlasGlossaryTerm.USER_DESCRIPTION]
  )
  ```

### QOL Improvements

- Added an [OSV](https://osv.dev) vulnerability-scan workflow job to GitHub Actions.

## 3.1.2 (December 31, 2024)

### New features

- Enabled the use of the `Retry-After` header for handling rate-limit retries.
- Added support for [OpenTelemetry](https://opentelemetry.io) logging in custom packages.
- Added `creator()` methods for Insights assets (`Collection`, `Folder`, `Query`).
- Added support for the following new connector types:
  - ANAPLAN
  - AWS_ECS
  - AWS_BATCH
  - AWS_LAMBDA
  - AWS_SAGEMAKER
  - DATAVERSE

### QOL improvements

- Generated the latest typedef models.
- Upgraded `jinja2` from `3.1.4` to `3.1.5` to address a security vulnerability.
- Fixed pagination assertions in `AuditSearch` unit tests and integration tests.

## 3.1.1 (December 26, 2024)

### Bug fixes

- Fixed a `ValidationError` caused by the private field (`_user_id`) in the `AtlanClient` constructor.

## 3.1.0 (December 26, 2024)

### New features

- Added support for automatic token refresh and retrying API requests upon receiving a `401` (Unauthorized) response.

  To enable this feature, the following constants must be configured:

  **1. Environment variables:**
  For regenerating the bearer access token:
  - `CLIENT_ID` (string)
  - `CLIENT_SECRET` (string)

  **2. Update `AtlanClient` field:**
  - `_user_id` (string, default: `None`): The unique identifier (GUID) of the user that the client impersonates.

  **Example:**
  ```python
  client = AtlanClient()
  client._user_id = "962c8f78-98a7-908f-9ec2-9e5b7ee7a09f"
  ```

### Breaking changes

- Introduced a new pagination approach in `SearchLogClient.search()` called **search log bulk search** (disabled by default). The SDK switches to this search operation automatically if results exceed a predefined threshold (e.g: `10,000` results). Alternatively, users can enable bulk search explicitly by setting `bulk=True` in `SearchLogClient.search()`. This breaking change affects searches that return more than `10,000` results — either the results will now be sorted differently or an error will be thrown.

- `SearchLogClient.search()` method will now raise an `InvalidRequestError` exception in the following scenarios:
  - when bulk search is enabled (`bulk=True`) and any user-specified sorting options are included in the search request.
  - when bulk search is disabled (`bulk=False`), the number of results exceeds the predefined threshold (e.g: `10,000` assets), and any user-specified sorting options are found in the search request.

  _This is because the bulk search approach for search logs ignores user-specified sorting and instead reorders results by the `createdAt` timestamps of log entries to efficiently handle large volumes of search logs._

## 3.0.0 (December 13, 2024)

### New features

- Added a new connector type (`BIGID`).
- Added support for the following options in `Batch` operations:
  - `update_only`: bool (default: `False`)
  - `track`: bool (default: `False`)
  - `case_insensitive`: bool (default: `False`)
  - `table_view_agnostic`: bool (default: `False`)
  - `creation_handling`: `AssetCreationHandling` (default: `AssetCreationHandling.FULL`)
- Added default timeouts (`read`, `connect`) to `AtlanClient`:
  - Total retries: 5
  - `AtlanClient.connect_timeout`: float (default: 30.0 seconds)
  - `AtlanClient.read_timeout`: float (default: 120.0 seconds)
- Added support for a new `parent_type` (`SnowflakeDynamicTable`) in `Column.creator()`.
- Added exposure for source-specific custom attributes (e.g: `Asset.custom_attributes`).
- Added handling for `error_cause` and `backend_error_id` in `ErrorInfo`.

### Bug fixes

- Fixed the generator to correctly handle the naming of the `Asset.DOMAIN_GUIDS` keyword search field.
- Fixed an issue where search pages (`IndexSearchResults` and `AuditSearchResults`) could overrun when the total results are just under the `_MASS_EXTRACT_THRESHOLD`.
- Fixed an issue with timestamp paging returning incomplete results when searching with a small page size (e.g: `2`) and assets with the same creation time.

### QOL improvements

- Generated the latest typedef models.
- Added the `@init_guid` decorator to the `updater()` method of assets to ensure that GUIDs are properly initialized and resolved in batch operations.
- Removed `type_name` validation from `Table`, `View`, and `Materialised View` to make them configurable when running `Batch` operations with `table_view_agnostic=True`.
- Removed deprecated `AssetClient.get_lineage()` integration tests.
- Updated integration test asset constants to align with the new tenant setup.

### Deprecated features

- Removed the deprecated `AssetClient.get_lineage()` method, which is slower and will no longer receive enhancements. Use the `AssetClient.get_lineage_list()` operation instead.

### Breaking changes

- `Batch` now accepts `AtlanClient` as the first parameter, replacing the previous use of `AssetClient`.

- Introduced a new pagination approach in `AuditClient.search()` called **audit bulk search** (disabled by default). The SDK switches to this search operation automatically if results exceed a predefined threshold (e.g: `10,000` results). Alternatively, users can enable bulk search explicitly by setting `bulk=True` in `AuditClient.search()`. This breaking change affects searches that return more than `10,000` results — either the results will now be sorted differently or an error will be thrown.

- `AuditClient.search()` method will now raise an `InvalidRequestError` exception in the following scenarios:
  - when bulk search is enabled (`bulk=True`) and any user-specified sorting options are included in the search request.
  - when bulk search is disabled (`bulk=False`), the number of results exceeds the predefined threshold (e.g: `10,000` assets), and any user-specified sorting options are found in the search request.

  _This is because the audit bulk search approach ignores user-specified sorting and instead reorders results based on the creation timestamps of assets to handle large volumes of assets efficiently._

## 2.7.0 (December 4, 2024)

### QOL improvements

- Renamed `CredentialResponseList` to `CredentialListResponse` to ensure consistent response model naming.
- Updated handling for (`{"records": null}`) in the response to populate the model with an empty list (`[]`) instead of `None`, which is non-iterable.
- Updated `level` and `connection` fields to use `Optional[Union[Dict[str, Any], str]]`, as they can be strings, preventing Pydantic validation errors, eg:

  ```
  "level": "user",
  "connection": "default/bigquery/1234567890"
  ```

## 2.6.2 (December 3, 2024)

### QOL improvements

- Added Column projection support to `group.get_all()`.
- Added retry on http code 429.

## 2.6.1 (November 25, 2024)

### New features

- Added the `"x-atlan-client-origin": "product_sdk"` header to the `AtlanClient`.
- Added a method to retrieve all credentials: `CredentialClient.get_all()`.

### QOL improvements

- Generated the latest typedef models.
  - Application typedef changes:
    - Replaced the `Application` supertype with `App`.
    - Renamed `ApplicationContainer` to `Application`.
    - Changed the level of the relationship from `Catalog` to `Asset`.
    - Updated the name of the de-normalized attribute.
- Added a wait to the integration test fixtures (atlan_tag_test, custom_metadata_test).

## 2.6.0 (November 21, 2024)

### New features

- Added the `creator()` method for `ApplicationContainer`.
- Added a new connector type (`APPLICATION`).
- Added support for a new method (`s3()`) in `TableauCrawler` to fetch metadata directly from S3 bucket extractions.
- Added support for the following CSA custom packages:
  - Asset import
  - Asset export (basic)
  - Relational assets builder
- Added `username` and `extras` fields to the `CredentialResponse` model.

### Bug fixes

- Fixed `Batch._track()` method to handle `AtlasGlossaryTerm` assets correctly.

### QOL improvements

- Fixed several issues related to tag deletion during integration test cleanups (e.g: `purpose_test`, `test_task_client`, and `suggestions_test`).
- Updated various response models in `pyatlan.model.workflow` to use `Optional` fields, ensuring complete capture of API response results.
- Removed `@validate_arguments` from `@overload` methods (`WorkflowClient`: `rerun()`, `add_schedule()`, `remove_schedule()`), as it unintentionally typecast arguments to different types, causing unexpected behavior when model fields were optional. Instead, replaced it with the utility function `validate_type()` to validate argument types in `@overload`ed methods.

## 2.5.8 (November 13, 2024)

### Bug fixes

- Fixed a typo in the IBM_DB2 connector type.

## 2.5.7 (November 13, 2024)

### New features

- Added support for new connector types:
  - IBM_DB2
  - SAP_GIGYA
  - SAP_HYBRIS
  - TREASURE_DATA
  - APACHE_PULSAR
  - ADOBE_TARGET
  - AZURE_ACTIVE_DIRECTORY
  - ADOBE_EXPERIENCE_MANAGER

### QOL improvements

- Generated the latest typedef models.

### Bug fixes

- Fixed generator to handle new core assets in the typedefs.

## 2.5.6 (October 24, 2024)

### Bug fixes

- Fixed `Readme.creator()` to use `asset.trim_to_reference()` instead of sending the complete `asset`, which was somehow breaking backend parsing for related assets.

### QOL improvements

- Generated the latest typedef models.

## 2.5.5 (October 23, 2024)

### Bug fixes

- Fixed `AtlanError` message for unescaped curly braces (`{}`) in `response.text`.

## 2.5.4 (October 22, 2024)

### New features

- Implemented cache management for `Connection` and `SourceTag`.
- Added support for assigning a `SourceTag` with a `value` to an asset.

## 2.5.3 (October 16, 2024)

### Bug fixes

- Added the missing `FluentLineage.includes_on_relations` method and the `LineageListRequest.relation_attributes` field.

### New features

- Added a new connector type (`INRIVER`).
- Added `APIObject`, `APIQuery`, and `APIField` assets, along with their `creator()` methods.
- Added a boolean field `immediate_neighbors` to the `LineageListRequest`, which allows users to include immediate neighbors of the starting `asset` in the response (`True`), or exclude them (`False` - default).

## 2.5.2 (October 11, 2024)

### Bug fixes

- Updated the type of `TableauDatasource.upstream_tables` and `upstream_datasources` to `Optional[List[Dict[str, Optional[str]]]]` to prevent Pydantic validation errors when a dict `value` in this field is mapped to `None`.

### QOL improvements

- Generated the latest typedef models.

## 2.5.1 (October 9, 2024)

### Bug fixes

- Fixed `ModuleNotFoundError` caused by the missing `PyYAML` dependency in the project’s `requirements.txt`.

## 2.5.0 (October 8, 2024)

### New features (experimental)

- Added support for sending `OpenLineage` events.
- Added support for serialization/deserialization (serde) of the `DataContract` spec.

### Breaking changes

- Updated the `ContractClient.generate_initial_spec()` method to directly accept the `asset` for which the initial contract spec is generated, similar to the Java SDK. Previously, you needed to pass the keyword arguments `asset_type` and `asset_qualified_name`.


## 2.4.8 (October 5, 2024)

### New features

- Added new connector types (`IICS`, `ABINITIO`, `SAP_S4_HANA`).

## 2.4.7 (September 30, 2024)

### New features

- Added `CompoundQuery.tagged_with_value()` method to search
for source-synced tags by assigned value eg: Snowflake tags.
- Added `ContractClient.generate_initial_spec()` method to generate
the initial contract specification for the provided asset `typeName` and `qualifiedName`.

### Experimental

- Added initial support for `Span`, `SpanNear`, `SpanWithin`, and `SpanTerm` queries for textual fields.

## 2.4.6 (September 24, 2024)

### New features

- Added a new connector type `MODEL` for data modeling assets.

### QOL improvements

- Generated latest typedefs models (`model`, `anomalo`, `powerbi`).

## 2.4.5 (September 18, 2024)

### New Features

- Added `username` property to the `ApiToken` model.
- Added functions for validating custom package files to `pyatlan.test_utils`.
- Added an optional parameter `asset_selection` to `DataProduct.updater()`, allowing users to update assets within the data product.
- Added the `DataProductsAssetsDSL.get_asset_selection()` method, which returns the asset selection DSL string for a data product based on the specified `IndexSearchRequest`.

### Bug Fixes

- Fixed pagination issues in `AuditSearchResults`.
- Fixed multipart form handling for `AtlanTag` image uploads.

## 2.4.4 (September 10, 2024)

### New features

- Added a new function (`pkg.utils.set_package_headers`) to configure the `AtlanClient` with the required custom package headers using environment variables.

### Bug fixes

- Fixed `pkg.utils.validate_connection()` to include other types (`Connection`, `Dict`).

### QOL improvements

- Moved `nanoid` from the `requirements-dev.txt`.
- Moved common test functions to a separate `test_utils` package.

## 2.4.3 (September 04, 2024)

### New features

- Added a new connector type `DM` for data modeling assets.

## 2.4.2 (August 28, 2024)

### New features

- Added support for `GCS` presigned URL file uploads.
- Added the `find_run_by_id` method to the `WorkflowClient`.

### QOL improvements

- Generated the latest typedef models.
- Added `pyatlan` to the Conda packages.

### Bug fixes

- Fixed the interpretation of `text-only` indexed fields for assets.

## 2.4.1 (August 12, 2024)

### New features

- Added support for `Azure Blob` presigned URL file uploads.

### QOL improvements

- Replaced `pyatlan.utils.HTTPStatus` with the standard library `http.HTTPStatus` (introduced in **Python 3.4**).

## 2.4.0 (August 06, 2024)

### QOL improvements

- **Implemented lazy imports for `pyatlan.model.assets`:** This change reduces the import time for assets by deferring the import of modules until they are actually needed. As a result, users will experience faster startup times and reduced memory usage when working with assets.

### Bug fixes

- Fixed missing check for `asset.guid` in the `Readme.creator()` method, which caused issues with updating `Readme` assets when the asset was passed using `Asset.ref_by_qualified_name()` instead of `Asset.ref_by_guid()`. The SDK now throws a `ValueError` if `asset.guid` is missing in the `Readme.creator()` method to prevent these issues.

## 2.3.3 (July 23, 2024)

### New features

- Added support for nested aggregations.
- Added support for trident suggestions.
- Added `Superset` assets `creator()` methods.

## 2.3.2 (July 16, 2024)

### New features

- Added a new ELT connector type: `PREFECT`.

### Bug fixes

- Fixed the `ErrorCode.CM_ATTR_NOT_FOUND_BY_NAME` error message in `CustomMetadataCache.get_attr_id_for_name()`, which was receiving the wrong parameters.

## 2.3.1 (July 9, 2024)

### New features

- Added `WorkflowClient.find_by_id()` method for workflow search/retrieval.

### OQL improvements

- Made `AtlanError` messages more descriptive.

## 2.3.0 (June 19, 2024)

### Breaking changes

- Introduced a new pagination approach in `AssetClient.search()` and `FluentSearch.execute()` called **bulk search** (disabled by default). It minimizes system impact when handling large result sets. The SDK switches to this search operation automatically if results exceed a predefined threshold (i.e: `100,000` results). Alternatively, users can enable bulk search explicitly by setting `bulk=True` in `AssetClient.search()` or `FluentSearch.execute()`. The breaking change is in regards to searches that return more than `100,000` results — either the results will now be sorted differently or an error will be thrown.

- The `AssetClient.search()` and `FluentSearch.execute()` methods will now raise an exception (`InvalidRequestError`) in the following scenarios:

  - when bulk search is enabled (`bulk=True`) and any user-specified sorting options are found in the search request.

  - when bulk search is disabled (`bulk=False`), the number of results exceeds the predefined threshold (i.e: `100,000` assets), and any user-specified sorting options are found in the search request.

  _This is because the bulk search approach ignores user-specified sorting and instead reorders the results based on the creation timestamps of assets to handle large numbers of assets efficiently._

### QOL improvements

- Pinned `urllib3>=1.26.0,<3` and moved `networkx` to the dev requirements to avoid potential version mismatches.

## 2.2.4 (June 11, 2024)

### New features

- Added the `creator()` method for `DataContract`.
- Overloaded the `creator()` methods to explicitly handle ancestral `qualified_name` and `name`. This enhancement allows users to create objects when the ancestral asset name contains `/`, which `creator()` was previously unable to handle.

### QOL improvements

- Generated the latest typedef models.
- Updated the latest set of Phosphor icons in the `AtlanIcon` enum.

## 2.2.3 (June 4, 2024)

### New features

- Added thread lock to cache methods (`get_cache()` and `_refresh_cache()`).
- Added support for the `Connection Delete` package.
- Added support for custom metadata handling for data domains and products.

### QOL improvements

- Generated the latest typedef models (`DatabricksUnityCatalogTag`).
- Updated `AtlanError` to include `user_action` when throwing an exception.

## 2.2.2 (May 28, 2024)

### New features

- Added creator methods to `Kafka` and `AzureEventHub` assets.
- Added support for `includeClassificationNames` in `IndexSearchRequest`.
- Expanded the functionality of the `WorkflowClient` with new methods for scheduled query workflows.

### Bug fixes

- Fixed the `AtlasGlossaryCategory.trim_to_required()` method to use `updater()` instead of `create_for_modification()`.

### QOL improvements

- Generated the latest typedef models, including `Workflow` and `WorkflowRun`.
- Updated the SSO provider in `test_sso_client.py` to check for `JUMPCLOUD` instead of `AZURE_AD`.
- Updated the Pydantic dependency (`pydantic>=2.0.0,<3.0.0`) to provide more flexibility for users.

## 2.2.1 (May 21, 2024)

### New features

- Added `find_domain_by_name()` and `find_product_by_name` methods to `AssetClient` to find data mesh objects by their human-readable names.
- Modified the type of the `attributes` and `related_attributes` keyword parameters in `asset.get_hierarchy()` to accept `str` in addition to `AtlanField`.

### Bug fixes

- Fixed issues with data domain and data product creation.
- Added the `AtlanConnectorType.get_connector_name()` method, which handles connection qualified name validation for assets and returns the connector name. Previously, this logic was repeated in most Asset `creator()` methods, which sometimes resulted in the `connector_name` field being mistakenly omitted (this bug was found in the `File` asset).

### QOL improvements

- Added data mesh integration tests.
- Added a fallback delete mechanism for API tokens in tests.
- Generated the latest typedef models, such as `Cognos`, `Stakeholder`, etc.

## 2.2.0 (May 15, 2024)

### Breaking changes

- Fixes typos and docstrings in IAM role methods of the `DynamoDB` and `PostgreSQL` crawlers. Previously, the method name was `iam_user_role_auth` instead of `iam_role_auth`.

## 2.1.9 (May 14, 2024)

### New features

- Added support for `BigQuery`, `DynamoDB` and `Postgres` crawlers.

### Bug fixes

- Fixed logic in the `Referenceable` model to determine the correct subtype.

### QOL improvements

- Utilized generated atlas core enums in the import template.
- Added a shell script (`pyatlan-formatter`) capable of handling code formatting for both tracked and untracked files.

## 2.1.8 (May 8, 2024)

### New features

- Added support for `Snowflake Miner`.
- Introduced new connector types: `COGNITE`, `SYNDIGO`, `NETEZZA`, and `AZURE_SERVICE_BUS`.
- Added an optional `depth` field to indicate the asset's depth within lineage in the `Referenceable` model.
- Added an optional parameter (`related_attributes`) to the `get_hierarchy()` method, allowing users to specify a list of attributes to retrieve for each related asset in the hierarchy.
- Expanded the functionality of the `WorkflowClient` with new methods for scheduling, stopping, and deleting workflows.

### QOL improvements

- Upgraded `jinja2` from `3.1.3` to `3.1.4` to address a security vulnerability: [GHSA-h75v-3vvj-5mfj](https://github.com/advisories/GHSA-h75v-3vvj-5mfj).

## 2.1.7 (April 30, 2024)

### New features

- Adds `FileClient` for uploading and downloading files from Atlan's tenant object storage via presigned URLs.

## 2.1.6 (April 23, 2024)

### New features

- Adds support for SSO group mapping.
- Adds `AtlanSSO` to enumerate the options for supported Atlan's SSO providers.

## 2.1.5 (April 18, 2024)

### Bug fixes

- Fixed an issue where explicit assignment of `None` to asset attributes in `BulkRequest` within the `client.asset.save()` method resulted in exclusion from the request payload.
- Fixed issues with **multiple remove/append relationships** where values were being overwritten inside append/remove dictionaries.

## 2.1.4 (April 16, 2024)

### New features

- Adds `AssetFilterGroup` enum for persona personalization.
- Adds creator method to `AirflowDag` and `AirflowTask` assets.

### Bug fixes

- Fixes `qualified_name` population for related entities in `Glossary` objects.
- Fixes logic for removing parent category relationship in `AtlasGlossaryCategory`.
- Fixes lineage list performance by considering the `hasMore` attribute to optimize lineage paging.

### QOL improvements

- Updates `networkx >= 3.1` and bumped development requirements to the latest version.

## 2.1.3 (April 3, 2024)

### Bug fixes

- Fixes field alias typo for `restrict_propagation_through_hierarchy` in the `AtlanTag` model.
- Fixes issues with `AtlanEvent` model deserialization. Previously, it threw a `ValidationError` for `ENTITY_DELETE`, `CLASSIFICATION_ADD`, and `CLASSIFICATION_DELETE` events.

## 2.1.2 (April 2, 2024)

### New features

- Adds `atlanMetadata` icon to `AtlanIcon` enum.
- Adds support for `restrict_propagation_through_hierarchy` to the `AtlanTag` model.
- Adds latest typedef models **(cube, cognite, spark, sparkjob, and multi-dimensional dataset)**.
- Adds support to limit the applicability of custom metadata attributes directly through the `AttributeDef.create()` method.

### Bug Fixes

- Fixes an issue with direct model deserialization of structs.

## 2.1.1 (March 28, 2024)

### Bug Fixes

- Fixes an issue where certain structs could cause errors during marshalling and unmarshalling.
- Fixes `Process.generate_qualified_name()` method to ensure that the generated process `qualified_name` includes the connection prefix. Previously, it only used the MD5 hash as the `qualified_name` for the `Process`.

## 2.1.0 (March 27, 2024)

### New features

- Adds `EnumDef.update()` method to update an existing enumeration with more valid values.
- Adds `get_by_name()` method to the `TypeDefClient` to retrieve a specific type definition from Atlan.
- Adds `propagation_only_through_lineage` field to the `AtlanTag`, allowing users to specify that
propagation will only occur downstream in the lineage and not within the hierarchy.

### Breaking changes

- `EnumCache.get_by_name()` method now raises `NotFoundError` if the enumeration with the given name does not exist. Previously the method would only return `None` in case the enumeration did not exist, rather than raising an error.

### QOL improvements

- Bump pre-commit black from `23.7.0` to `24.3.0` (Fixes [CVE security vulnerability](https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2024-21503)).

## 2.0.4 (March 22, 2024)

### New features

- Adds model field (`__atlan_extra__`) to the `AtlanObject` model to accommodate extra properties from API responses.
- Adds an optional parameter (`return_info`) to the user creation method, allowing control over whether to return the list of created users or not.
- Adds an optional parameter (`exclude_users`) to the `SearchLogRequest` methods for excluding specific user data from search log results.

### Bug fixes

- Fixes issues related to creating data domains and data products.

### QOL improvements

- Bump black from `23.7.0` to `24.3.0` (Fixes [CVE security vulnerability](https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2024-21503)).

## 2.0.3 (March 15, 2024)

### New features

- **(Experimental)** Adds support for `Sigma` and `SQL Server` crawlers.
- Adds AWS IoT connector types: `AWS_SITE_WISE` and `AWS_GREENGRASS`.
- Adds credential widget to package toolkit.

### Bug fixes

- Fixes `to_dict` method handling of `Range` queries for `datetime` attributes.
- Fixes `UserClient` create method throwing 500 (internal server error).

### QOL improvements

- Sets `asset_type` keyword argument default value to `Asset` in `get_by_guid` and `retrieve_minimal` methods.

## 2.0.2 (March 5, 2024)

### New features

- **(Experimental)** Adds the ability to search the background tasks queue.
- Adds the missing connector type for `MICROSTRATEGY`.

### Bug fixes

- Replaces spaces with underscores in generated enums.

## 2.0.1 (February 27, 2024)

### New features

- Adds pagination to listings for **user**, **group**, and **membership**.

### Bug fixes

- Fixes `LineageListResults` not including custom metadata.
- Avoid throwing `ValidationError` when aggregating an index search with no results.

### QOL improvements

- Improves the tests report experience by using pytest plugins (`pytest-sugar` and `pytest-timer`).

## 2.0.0 (February 22, 2024)

### New features

- Migrates to Pydantic 2.6
- Allows use of Python 3.8 as a minimum version supported by the SDK
- Adds file connector as a new connector type
- Adds Essbase connector as a new connector type
- Allows fluent lineage conditions to be configured for combining (AND) or any (OR)
- Adds an option to rerun workflows idempotently

### Breaking changes

Since this is a new major release, there are some breaking changes:

- The move to Pydantic 2.6 means we have a dependency on a different major release of this library than previously. We do not believe it will impact your use of any of our methods; however, it could impact you if you rely on Pydantic v1 elsewhere in your codebase and are not yet ready to move to Pydantic v2 yourself.
- The `find_personas_by_name`, `find_purposes_by_name` and `find_connections_by_name` methods previously returned an empty list if no requested object were found, despite being documented as throwing a `NotFoundError` in such cases. We have now made this consistent with the other `find_.._by_name` methods, so if a requested object does not found it will now throw a `NotFoundError`.
- To create consistency across our SDKs, we are also introducing new asset creation and modification methods. This is to reserve the use of verbs in a method name for server-side interactions (like `save`, `find`, etc) and instead use non-verbs to indicate that you are creating an instance of an object purely in-memory — but that it still needs to be acted upon (`save`d) in order to be persisted to Atlan. These offer identical functionality to the prior methods, and we will keep the prior methods around until at least the next major release but simply mark them as deprecated. (So no immediate change to any existing code should be needed.)

    | Old method | New method |
    |---|---|
    | `create()` | `creator()` |
    | `create_for_modification()` | `updater()` |

## 1.9.4 (February 13, 2024)

### New features

- Adds a way to specify replace, append, or removal of individual relationships during `save` operations

### Bug fixes

- Adds missing changes to code generation templates
- Corrects an erroneous use of the inverse operator

## 1.9.3 (January 30, 2024)

### Bug fixes

- raise exception if tag in query in FluentSearch does not exist
- (Experimental) fix defects custom package generation

### QOL Improvements

- add api to assign a token to a purpose
- add api to enable running sql queries

## 1.9.2 (January 24, 2024)

### Bug fixes

- Fix problem with AssetUpdatePayload not being returned from AtlanEvent
- Fix problem with sort order for AuditSearchRequest

## 1.9.1 (January 23, 2024)

### QOL Improvements

- Make `aws_arn` optional in S3 models
- Change default S3 file in package toolkit
- Update docstrings in package toolkit
- Add workflow `update` method
- Adds QA checks (black, flake8, mypy) to the CI
- Bump jinja2 from 3.1.2 to 3.1.3
- Add missing regex and wildcard methods to the `KeywordField`

### Bug fixes

- Make sure the name of the python file generated by the package toolkit does not contain any special characters
- Correct the description of fields in the `AttributeDef` model


## 1.9.0 (January 16, 2024)

### New Features

- Add ability to update certificate, announcement for GlossaryTerm and GlossaryCategory
- Add `create` method for `ColumnProcess`
- Always include sort by `GUID` as final criteria in `IndexSearch`
- (Experimental) Add classes to support custom package generation

### QOL Improvements

- Add an additional parameter to `create` method of `ADLSObject`
- Add type checking to AtlanClient save and other methods

## 1.8.4 (January 4, 2024)

### New Features

- Add credential management

### Bug fixes

- Fix workflow client monitor method
- Fix problem where Python SDK was not initializing applicableConnections and applicableGlossaries when creating new custom metadata sets
- Fix handling of archived custom metadata attributes

## 1.8.3 (December 27, 2023)

### New Features

- (Experimental) Define and run new workflows
- Added support for retrieval of `SourceTagAttachements` in `AtlanTags`

### Bug fixes

- Fix duplicate initial GUID on Windows. This was due to a lack of support for nano-second accuracy in timestamps in Python on Windows. Code has been changed to rectify this problem.

## 1.8.2 (December 21, 2023)

### New Features

### Bug fixes

- Fix problem parsing assets that contain SourceTagAttachements

### QOL improvements


## 1.8.1 (December 19, 2023)

### New Features

- Add support for search log access

### Breaking changes

### Bug fixes

- Modify constants like Referenceable.TYPE_NAME to use InternalKeywordField, InternalKeywordTextField or InternalNumericField
- Fix groups in purpose policies and remove data mesh slugs and abbreviations
- Added missing "templates" field to the WorkflowSpec model

### QOL improvements

- Change qualifiedName generation on Link to be deterministic

## 1.8.0 (December 13, 2023)

### New features

### Breaking changes

- `Asset.create_for_modification` now raises an exception. The `create_for_modification` method should not be invoked on the `Asset` class. Instead `create_for_modification` should be invoked on the class of the asset to be created, for example `Table.create_for_modification`.

### Bug fixes

- An exception was being raised when `client.asset.replace_custom_metadata` was invoked on a custom metadata set that contained archived attributes.
- The `ADLSObject.create()` was missing the parameter for `adls_account_qualified_name`. This was causing the `ADSLObject` to be displayed incorrectly in the UI.

### QOL improvements


## 1.7.0 (December 1, 2023)

### New features

- (Experimental) Adds initial support for data domains and data products

### Breaking changes

- Disables the `delete_by_guid()` and restore methods for categories. Categories can currently only be purged (hard-deleted), so even when previously calling the `delete_by_guid()` method the backend actually translated this to a hard-delete (purge). (Which also means there is no way to restore a category.) We have therefore opted to disable the `delete_by_guid()` and restore methods — only for categories — to make it clearer that categories can only be purged (hard-deleted).

### Bug fixes

- Fixes error shown on retry overrun to show the underlying error, not the retry limit
- Fixes a message that was logged as an error but was purely informational (debug-level)

### QOL improvements

- Adds a unique request ID to every request, and includes in logging
- Logging is now done in both plain text and JSON
- Bundles a reasonable default logging configuration to the base container image

### Background changes

- Consolidates Python-based code repositories into a single place

## 1.6.4 (November 21, 2023)

### Deprecations
* None
### Bug Fixes
* Fix problem with source_read_top_user_record_list not being read correctly in index search
* Correct problem with AuditSearch mis-identifying asset as deleted tag
### QOL Improvm
* Update enums with latest changes to persona preferences
* Bundle logging.conf file in Docker image
* Add ability to set icon during glossary creation

## 1.6.3 (November 14, 2023)

* Changed to support new custom metadata payload
* Add methods to manage Preset Assets
* Add in-cluster optimisation to use services URLs directly
* Changes the logic for generating Google Data Studio asset qualifiedNames

## 1.6.2 (November 7, 2023)

* Add filter to redact HTTP authorization header
* Add ability to traverse glossary hierarchy
* Publish docker image on release

## 1.6.1 (November 3, 2023)

* PARENT_CATEGORY of Glossary had the wrong atlan_field_name

## 1.6.0 (November 1, 2023)

* Rename exclude_classifications to exclude_atlan_tags in FluentLineage

## 1.5.0 (October 31, 2023)

* Add FluentLineage an abstraction mechanism, to simplify the most common lineage requests against Atlan (removing the need to understand the guts of Elastic).
* Add support for management of Google Data Studio assets

## 1.4.2 (October 30, 2023)

* Fix bug causing token_client.get_by_id to fail when when deprecated personas section present

## 1.4.1 (October 25, 2023)

* Fix bug causing client.user.add_as_admin to fail
* Add create methods for GCSBucket and GCSObject

## 1.4.0 (October 19, 2023)

* Create new helper client classes:
  * AdminClient
  * AssetClient
  * Add property atlan_tag_names to assets to provide human readable versions of classificationNames
  * Added Batch class for managing bulk updates in batches.

## 1.3.3 (October 12, 2023)

* Create new helper client classes:
  *  GroupClient
  *  RoleClient
  *  TokenClient
  *  TypeDefClient
  *  UserClient
* Add custom pre-commit hook
* Additional optional parameters for index searc
* Add capability to manage API Assets


## 1.3.2 (October 10, 2023)

* Fix bug preventing classifications with unexpected attributes from being read

## 1.3.1 (October 6, 2023)

* Fix bug preventing custom metadata from being read that were written with locked bug

## 1.3.0 (October 5, 2023)

* Correct issue with incorrect type on purpose_atlan_tags of Purpose
* Added support for searching audit logs
* Fixed bug preventing creation of locked custom metadata
* Fully-validate connection parameters on save
* Adds emojis and icons for both tags and custom metadata
* Fix bug on validating (internal) group names
* Manage ADLS assests
* Makes all options in AttributeDefs optional

## 1.2.0 (September 26, 2023)

* Add capability to do bucket and metric aggregations
* Mark lineage method (and others) deprecated, including DeprecationWarning

## 1.1.0 (September 20, 2023)

* Add capability to find and re-run workflows
* Fix Range query with 0 not generating expected request
* Adds Matillion, MongoDB, and time granularity for custom metadata
* Generates Sphinx documentation for the SDK
* Adds an API token check as part of cache refreshes

## 1.0.0 (September 13, 2023)
* Initial production release
* Adds column projections for listing users
* Renamed register_client method to set_default_client from AtlanClient
* Removed reset_default_client from AtlanClient

## 0.8.0 (September 11, 2023)
* Improve retry strategy for HTTP
* Adds generators for developer portal docs
* Adds create helper for Links
* Fixes bug in adding API tokens as admins and viewers
* Removes ability to activate and deactivate users
* Refactored Exception hierarchy

## 0.7.0 (September 4, 2023)
* Upgrade to pydantic 1.10.12

## 0.6.3 (August 31, 2023)
* Add ability to get, create and modify API tokens

## 0.6.2 (August 24, 2023)
* Fix bug in generating field class vars

## 0.6.1 (August 23, 2023)
* Adds a new FluentSearch approach to searching (no need to know Elastic anymore)
* Changes default inclusion operation for Bool queries to filter rather than must (for efficiency)

## 0.6.0 (August 17, 2023)
* Added:
  * Adds new purpose policy permissions (attaching / detaching terms from assets)
  * Fixes type for decentralized_role
  * Documents various public methods in client, caches, and event-handling classes
  * Add code generation for enums from enum-defs
  * split assets into multiple modules
  * Initial support for an X-Atlan-Request-Id header
* Breaking changes
  * find_category_by_name[_fast] now return a list of categories rather than a single category, since categories are not unique by name within a glossary.

## 0.5.0 (August 10, 2023)
* Added:
  * Bulk delete
  * Restore
  * Search by certificate
  * Activate / deactivate users
  * Search by any of multiple types
* regenerates model to include latest assets and structs
* Breaking change:
  * S3_bucket_qualified_name is now required for S3 object creation

## 0.4.3 (August 7, 2023)

* Add a creation-only semantic operation, and deprecate upsert in favour of save
*
## 0.4.2 (July 25, 2023)

* Add Alteyx to AtlanConnectorType enums

## 0.4.1 (July 24, 2023)

* Correct issue where custom metadata wasn't being returned after the first page in mult-page index search
* Avoid double zipping when producing lambda layer in build
* Correct problem with indexing error
* Removed redundant code related to custom metadata
* Add handling for log events
* Add constant for archived asset retrieval

## 0.4.0 (July 5, 2023)

* Pin dependency on pydantic==1.10.8
* Facilitate inclusion of custom metadata in index search
* Change lineage list response into a lazy-fetching generator

## 0.3.0 (July 3, 2023)

* Renamed AtlanClassificationColor to AtlanTagColor

## 0.2.0 (June 30, 2023)

* Renamed all references to Classification and classification to Tag and tag (TagDef, Tag, tags, tagNames, etc):
* Add ability to create Files as an asset directly on Atlan.
* Add Event-handling abstractions
* Add the following methods to pyatlan.client.atlan.AtlanClient
  * find_glossary_by_name
  * find_category_fast_by_name
  * find_category_by_name
  * find_term_fast_by_name
  * find_term_by_name

## 0.1.3 (June 19, 2023)

* Adds new persona and purpose policy handling
* Added trim_to_required method to assets

## 0.1.2 (June 14, 2023)

* Set status to Active when Badge is created

## 0.1.1 (June 12, 2023)

* Adds icons and images for classifications
* Migrates to easier to use uniqueness generator for testing
* Refactor Integration Tests to work on any server
* Remove incorrect required field validations for Connection


## 0.1.0 (June 6, 2023)

* Fixes connection creation and associated validations
* Adds further connectors (with icons)
* Adds SQL parser endpoint
* Add Lineage list API and integration tests
* Fix bug on locking custom metadata
* Refactor syntax for accessing custom metadata properties

  Previously the custom metadata property `First Name` would have been accessed as follows:
    ````cm.first_name````

  This could cause confusion how the name was converted so now the following syntax will be used:

    ```cm['First Name']```
## 0.0.33 (May 24, 2023)

* Added convenience properties for relationship_attributes
* Rename convenience property for glossary terms from terms to assigned_terms

* Add create method to Badge
* Add integrations tests for CustomMetaData

## 0.0.31 (May 15, 2023)

* Added the following classes to support lineage retrieval
  * LineageRelation
  * DirectedPair
  * LineageGraph
  * LineageResponse
* Added the get_lineage method to AtlanClient
* Modify create_typdef in client to handle creating EnumDef
* Refresh caches on any SDK-driven creates or deletes

## 0.0.30 (May 11, 2023)

* Fix problem where custom metadata created via the SDK failed to show up in the UI

## 0.0.29 (May 11, 2023)

* Fix problem where custom metadata created via the SDK failed to show up in the UI

## 0.0.28 (May 9, 2023)

* Add find_connections_by_name to AtlanClient

## 0.0.27 (May 8, 2023)

* Add a create method to Process

## 0.0.26 (May 4, 2023)

* Add remove_terms method to AtlanClient
* Add append_terms method to AtlanClient
* Add replace_terms method to AtlanClient

## 0.0.25 (May 2, 2023)

* Update create method for Readme asset

## 0.0.24 (Apr 27, 2023)

* Fix broken link in README.md

## 0.0.23 (Apr 27, 2023)

* Renamed get_business_attributes method get_custom_metadata in Referenceable.Attributes
* Renamed set_business_attributes method set_custom_metadata in Referenceable.Attributes
* Renamed BusinessAttributes to CustomMetadata
* Add parameter overwrite_custom_metadata to AtlanClient.upsert
* Add get_custom_metadata to CustomMetadataCache
* Update create functions

## 0.0.22 (Apr 12, 2023)

* Added create method to Column
* Updated Schema.create method to add schema to Database schemas collection
* Updated Table.create method to add table to Schema tables collection

## 0.0.21 (Apr 11, 2023)

* Added relation_attributes parameter to IndexSearchRequest to specify the attributes to be included in each relationship that is included in the results of the search
* Set assets to empty list when no entities are returned by search
* Add classmethod ref_by_guid to Referencable
* Add classmethod ref_by_qualified_name to Referencable

## 0.0.20 (Apr 5, 2023)

* Corrected issue with from, size and track_total_hits in DSL object not being sent to server if the defaults weren't changed
* Added relation_attributes parameter to IndexSearchRequest to specify the attributes to be included in each relationship that is included in the results of the search

## 0.0.19 (Mar 30, 2023)

* Renamed environment variable used to specify the Atlan URL from ATLAN_HOST to ATLAN_BASE_URL
* Renamed initialization parameter in pyatlan.client.atlan.AtlanClient from host to base_url

## 0.0.18 (Mar 27, 2023)

* Added workflow to run code scan with [CodeQl](https://codeql.github.com/)
* Added workflow to upload to [pypi](https://pypi.org) on release

## 0.0.17 (Mar 22, 2023)

* Initial public release