New Features ✨

Ai

• feat(ai): add cache writes for gen_ai by @shellmayr in #5319
• feat(ai): add parse_data_uri function to parse a data URI by @constantinius in #5311

Other

• feat(asyncio): Add on-demand way to enable AsyncioIntegration by @sentrivana in #5288

  You can now enable the AsyncioIntegration on demand, after calling sentry_sdk.init(). This is useful in scenarios where you don't have
  the event loop running early on, or when you need to instrument multiple event loops.

importsentry_sdkfromsentry_sdk.integrations.asyncioimportenable_asyncio_integration# Initializing the SDK as early as possible, when there is no event loop yetsentry_sdk.init( ... # No AsyncioIntegration in explicitly provided `integrations` ) asyncdefmain(): enable_asyncio_integration() # instruments the current event loop# ...your code...

• feat(openai-agents): Inject propagation headers for HostedMCPTool by @alexander-alderman-webb in #5297
• feat(stdlib): Handle proxy tunnels in httlib integration by @sl0thentr0py in #5303
• feat: Support array types for logs and metrics attributes by @alexander-alderman-webb in #5314

Bug Fixes 🐛

Integrations

• fix(integrations): google genai report image inputs by @constantinius in #5337
• fix(integrations): google-genai: reworked gen_ai.request.messages extraction from parameters by @constantinius in #5275
• fix(integrations): pydantic-ai: properly format binary input message parts to be conformant with the gen_ai.request.messages structure by @constantinius in #5251
• fix(integrations): Anthropic: add content transformation for images and documents by @constantinius in #5276
• fix(integrations): langchain add multimodal content transformation functions for images, audio, and files by @constantinius in #5278

Litellm

• fix(litellm): fix gen_ai.request.messages to be as expected by @constantinius in #5255
• fix(litellm): Guard against module shadowing by @alexander-alderman-webb in #5249

Other

• fix(ai): redact message parts content of type blob by @constantinius in #5243
• fix(clickhouse): Guard against module shadowing by @alexander-alderman-webb in #5250
• fix(gql): Revert signature change of patched gql.Client.execute by @alexander-alderman-webb in #5289
• fix(grpc): Derive interception state from channel fields by @alexander-alderman-webb in #5302
• fix(pure-eval): Guard against module shadowing by @alexander-alderman-webb in #5252
• fix(ray): Guard against module shadowing by @alexander-alderman-webb in #5254
• fix(threading): Handle channels shadowing by @sentrivana in #5299
• fix(typer): Guard against module shadowing by @alexander-alderman-webb in #5253
• fix: Stop suppressing exception chains in AI integrations by @alexander-alderman-webb in #5309
• fix: Send client reports for span recorder overflow by @sentrivana in #5310

Documentation 📚

• docs(metrics): Remove experimental notice by @alexander-alderman-webb in #5304
• docs: Update Python versions banner in README by @sentrivana in #5287

Internal Changes 🔧

Fastmcp

• test(fastmcp): Narrow AttributeError try-except by @alexander-alderman-webb in #5339
• test(fastmcp): Stop accessing non-existent attribute by @alexander-alderman-webb in #5338

Release

• ci(release): Bump Craft version to fix issues by @BYK in #5305
• ci(release): Switch from action-prepare-release to Craft by @BYK in #5290

Other

• chore(gen_ai): add auto-enablement for google genai by @shellmayr in #5295
• chore(repo): Add Claude Code settings with basic permissions by @philipphofmann in #5342
• ci: 🤖 Update test matrix with new releases (01/19) by @github-actions in #5330
• ci: Add periodic AI integration tests by @alexander-alderman-webb in #5313
• chore: Use pull_request_target for changelog preview by @BYK in #5323
• chore: add unlabeled trigger to changelog-preview by @BYK in #5315
• chore: Add type for metric units by @sentrivana in #5312
• ci: Update tox and handle generic classifiers by @sentrivana in #5306

New Features ✨

• feat(api): Add Scope.set_attribute by @sentrivana in #5256

Bug Fixes 🐛

• fix(grpc): Gate third-party imports by @alexander-alderman-webb in #5246
• fix(opentelemetry): Gate third-party imports by @alexander-alderman-webb in #5247
• fix(ray): Keep variadic kwargs last in signatures by @alexander-alderman-webb in #5244
• fix(trytond): Gate third-party imports by @alexander-alderman-webb in #5245
• Fix openai count_tokens by @sl0thentr0py in #5281

Documentation 📚

• docs: Fix typo in comment by @sentrivana in #5280
• docs: Fix middleware_spans docstring by @sentrivana in #5279

Internal Changes 🔧

• ref(scope): Set global attrs on global scope by @sentrivana in #5259
• chore: Ignore type migration for scripts/ and tests/ in blame by @alexander-alderman-webb in #5284
• ref: Properly override parent func by @sentrivana in #5283
• ci: Allow to use Craft's new auto-versioning by @sentrivana in #5218
• ref: Deduplicate batchers by @sentrivana in #5263
• tests: Add dedicated transport format test for metrics, logs by @sentrivana in #5264
• ci: 🤖 Update test matrix with new releases (01/05) by @github-actions in #5273
• tests: General logs tests should use Sentry logs API by @sentrivana in #5262
• tests: Test preserialization of attributes by @sentrivana in #5260
• ci: Unpin Pydantic 1.x version in tests by @alexander-alderman-webb in #5261
• ref: Make logs, metrics go via scope by @sentrivana in #5213
• ci: Fix failing arq, fastapi tests on 3.7; update test matrix by @sentrivana in #5258

Middleware spans are now disabled by default in Django, Starlette and FastAPI integrations. Set the middleware_spans integration-level
option to capture individual spans per middleware layer. To record Django middleware spans, for example, configure as follows

importsentry_sdkfromsentry_sdk.integrations.djangoimportDjangoIntegrationsentry_sdk.init( dsn="<your-dsn>", integrations=[ DjangoIntegration(middleware_spans=True), ], )

New Features ✨

• feat(ai): add single message truncation by @shellmayr in #5079

• feat(django): Add span around Task.enqueue by @sentrivana in #5209

• feat(starlette): Set transaction name when middleware spans are disabled by @alexander-alderman-webb in #5223

• feat: Add "K_REVISION" to environment variable release check (exposed by cloud run) by @rpradal in #5222

Langgraph

• feat(langgraph): Response model attribute on invocation spans by @alexander-alderman-webb in #5212
• feat(langgraph): Usage attributes on invocation spans by @alexander-alderman-webb in #5211

OTLP

• feat(otlp): Optionally capture exceptions from otel's Span.record_exception api by @sl0thentr0py in #5235
• feat(otlp): Implement new Propagator.inject for OTLPIntegration by @sl0thentr0py in #5221

Bug Fixes 🐛

Integrations

• fix(django): Set active thread ID when middleware spans are disabled by @alexander-alderman-webb in #5220

• fix(integrations): openai-agents fixing the input messages structure which was wrapped too much in some cases by @constantinius in #5203

• fix(integrations): openai-agents fix multi-patching of get_model function by @constantinius in #5195

• fix(integrations): add values for pydantic-ai and openai-agents to _INTEGRATION_DEACTIVATES to prohibit double span creation by @constantinius in #5196

• fix(logs): Set span_id instead of sentry.trace.parent_span_id attribute by @sentrivana in #5241

• fix(logs, metrics): Gate metrics, logs user attributes behind send_default_pii by @sentrivana in #5240

• fix(pydantic-ai): Stop capturing internal exceptions by @alexander-alderman-webb in #5237

• fix(ray): Actor class decorator with arguments by @alexander-alderman-webb in #5230

• fix: Don't log internal exception for tornado user auth by @sl0thentr0py in #5208

• fix: Fix changelog config by @sentrivana in #5192

Internal Changes 🔧

• chore(django): Disable middleware spans by default by @alexander-alderman-webb in #5219

• chore(starlette): Disable middleware spans by default by @alexander-alderman-webb in #5224

• ci: Unpin Python version for LiteLLM tests by @alexander-alderman-webb in #5238

• ci: 🤖 Update test matrix with new releases (12/15) by @github-actions in #5229

• chore: Ignore type annotation migration in blame by @alexander-alderman-webb in #5234

• ref: Clean up get_active_propagation_context by @sl0thentr0py in #5217

• ref: Cleanup outgoing propagation_context logic by @sl0thentr0py in #5215

• ci: Pin Python version to at least 3.10 for LiteLLM by @alexander-alderman-webb in #5202

• test: Remove skipped test by @sentrivana in #5197

• Convert all remaining type annotations into the modern format by @zsol in #5239

• Convert sentry_sdk type annotations into the modern format by @zsol in #5206

Bug Fixes 🐛

• fix: Make PropagationContext.from_incoming_data always return a PropagationContext by @sl0thentr0py in #5186
• fix(integrations): anthropic set GEN_AI_OPERATION_NAME by @constantinius in #5185
• fix(spotlight): align behavior with SDK spec by @BYK in #5169
• fix(integrations): do not exit early when config is not passed as it is not required and prohibits setting gen_ai.request.messages by @constantinius in #5167
• fix(langchain): add gen_ai.response.model to chat spans by @shellmayr in #5159
• fix(integrations): add the system prompt to the gen_ai.request.messages attribute by @constantinius in #5161
• fix(ai): Handle Pydantic model classes in _normalize_data by @skalinchuk in #5143
• fix(openai-agents): Avoid double span exit on exception by @alexander-alderman-webb in #5174
• fix(openai-agents): Store invoke_agent span on agents.RunContextWrapper by @alexander-alderman-webb in #5165
• Add back span status by @sl0thentr0py in #5147

New Features ✨

• feat(integrations): openai-agents: add usage and response model reporting for chat and invoke_agent spans by @constantinius in #5157
• feat: Implement strict_trace_continuation by @sl0thentr0py in #5178
• feat(integration): pydantic-ai: properly report token usage and response model for invoke_agent spans by @constantinius in #5153
• feat(integrations): add support for embed_content methods in GoogleGenAI integration by @constantinius in #5128
• feat(logs): Record discarded log bytes by @alexander-alderman-webb in #5144
• feat: Add an initial changelog config by @sentrivana in #5145
• feat(django): Instrument database rollbacks by @alexander-alderman-webb in #5115
• feat(django): Instrument database commits by @alexander-alderman-webb in #5100
• feat(openai-agents): Truncate long messages by @alexander-alderman-webb in #5141
• Add org_id support by @sl0thentr0py in #5166

Deprecations

• Deprecate continue_from_headers by @sl0thentr0py in #5160

Build / dependencies / internal 🔧

• Remove unsupported SPANSTATUS.(ERROR|UNSET) by @sl0thentr0py in #5146
• Rename setup_otlp_exporter to setup_otlp_traces_exporter by @sl0thentr0py in #5142
• Simplify continue_trace to reuse propagation_context values by @sl0thentr0py in #5158
• Make PropagationContext hold baggage instead of dynamic_sampling_context by @sl0thentr0py in #5156
• Cleanup PropagationContext.from_incoming_data by @sl0thentr0py in #5155
• chore: Add commit_patterns to changelog config, remove auto-labeler by @sentrivana in #5176
• build(deps): bump actions/github-script from 7 to 8 by @dependabot in #5171
• build(deps): bump supercharge/redis-github-action from 1.8.1 to 2 by @dependabot in #5172
• ci: 🤖 Update test matrix with new releases (12/01) by @github-actions in #5173
• ci: Add auto-label GH action by @sentrivana in #5163
• ci: Split up Test AI workflow by @alexander-alderman-webb in #5148
• ci: Update test matrix with new releases (11/24) by @alexander-alderman-webb in #5139
• test: Import integrations with empty shadow modules by @alexander-alderman-webb in #5150
• Add deprecations to changelog categories by @sentrivana in #5162

Various fixes & improvements

• Preserve metadata on wrapped coroutines (#5105) by @alexander-alderman-webb
• Make imports defensive to avoid ModuleNotFoundError in Pydantic AI integration (#5135) by @alexander-alderman-webb
• Fix OpenAI agents integration mistakenly enabling itself (#5132) by @sentrivana
• Add instrumentation to embedding functions for various backends (#5120) by @constantinius
• Improve embeddings support for OpenAI (#5121) by @constantinius
• Enhance input handling for embeddings in LiteLLM integration (#5127) by @constantinius
• Expect exceptions when re-raised (#5125) by @alexander-alderman-webb
• Remove MagicMock from mocked ModelResponse (#5126) by @alexander-alderman-webb

Various fixes & improvements

• OTLPIntegration (#4877) by @sl0thentr0py

  Enable the new OTLP integration with the code snippet below, and your OpenTelemetry instrumentation will be automatically sent to Sentry's OTLP ingestion endpoint.

  importsentry_sdkfromsentry_sdk.integrations.otlpimportOTLPIntegrationsentry_sdk.init( dsn="<your-dsn>", # Add data like inputs and responses;# see https://docs.sentry.io/platforms/python/data-management/data-collected/ for more infosend_default_pii=True, integrations=[ OTLPIntegration(), ], )

  Under the hood, this will setup:

  • A SpanExporter that will automatically set up the OTLP ingestion endpoint from your DSN
  • A Propagator that ensures Distributed Tracing works
  • Trace/Span linking for all other Sentry events such as Errors, Logs, Crons and Metrics

  If you were using the SentrySpanProcessor before, we recommend migrating over to OTLPIntegration since it's a much simpler setup.

• feat(integrations): implement context management for invoke_agent spans (#5089) by @constantinius

• feat(loguru): Capture extra (#5096) by @sentrivana

• feat: Attach server.address to metrics (#5113) by @alexander-alderman-webb

• fix: Cast message and detail attributes before appending exception notes (#5114) by @alexander-alderman-webb

• fix(integrations): ensure that GEN_AI_AGENT_NAME is properly set for GEN_AI spans under an invoke_agent span (#5030) by @constantinius

• fix(logs): Update sentry.origin (#5112) by @sentrivana

• chore: Deprecate description truncation option for Redis spans (#5073) by @alexander-alderman-webb

• chore: Deprecate max_spans LangChain parameter (#5074) by @alexander-alderman-webb

• chore(toxgen): Check availability of pip and add detail to exceptions (#5076) by @alexander-alderman-webb

• chore: add MCP SDK Pydantic AI and OpenAI Agents to the list of auto enabled integrations (#5111) by @constantinius

• test: add tests for either FastMCP implementation (#5075) by @constantinius

• fix(ci): Re-enable skipped tests (#5104) by @sentrivana

• ci: 🤖 Update test matrix with new releases (11/17) (#5110) by @github-actions

• ci: Force coverage core ctrace for 3.14 (#5108) by @sl0thentr0py

Various fixes & improvements

• fix(openai): Check response text is present to avoid AttributeError (#5081) by @alexander-alderman-webb
• fix(pydantic-ai): Do not fail on new ToolManager._call_tool() parameters (#5084) by @alexander-alderman-webb
• tests(huggingface): Avoid None version (#5083) by @alexander-alderman-webb
• ci: Pin coverage version for 3.14 Django tests (#5088) by @alexander-alderman-webb
• chore: X handle update (#5078) by @cleptric
• fix(openai-agents): add input messages to errored spans as well (#5077) by @shellmayr
• fix: Add hard limit to log batcher (#5069) by @alexander-alderman-webb
• fix: Add hard limit to metrics batcher (#5068) by @alexander-alderman-webb
• fix(integrations): properly handle exceptions in tool calls (#5065) by @constantinius
• feat: non-experimental before_send_metric option (#5064) by @alexander-alderman-webb
• feat: non-experimental enable_metrics option (#5056) by @alexander-alderman-webb
• fix(integrations): properly distinguish between network.transport and mcp.transport (#5063) by @constantinius
• feat(integrations): add ability to auto-deactivate lower-level integrations based on map (#5052) by @shellmayr
• Fix NOT_GIVEN check in anthropic (#5058) by @sl0thentr0py
• ci: 🤖 Update test matrix with new releases (11/03) (#5054) by @github-actions
• Add external_propagation_context support (#5051) by @sl0thentr0py
• chore: Remove enable_metrics option (#5046) by @alexander-alderman-webb
• Allow new integration setup on the instance with config options (#5047) by @sl0thentr0py
• ci: Run integration tests on Python 3.14t (#4995) by @alexander-alderman-webb
• docs: Elaborate on Strawberry autodetection in changelog (#5039) by @sentrivana

Various fixes & improvements

• Pydantic AI integration (#4906) by @constantinius

  Enable the new Pydantic AI integration with the code snippet below, and you can use the Sentry AI dashboards to observe your AI calls:

  importsentry_sdkfromsentry_sdk.integrations.pydantic_aiimportPydanticAIIntegrationsentry_sdk.init( dsn="<your-dsn>", # Set traces_sample_rate to 1.0 to capture 100%# of transactions for tracing.traces_sample_rate=1.0, # Add data like inputs and responses;# see https://docs.sentry.io/platforms/python/data-management/data-collected/ for more infosend_default_pii=True, integrations=[ PydanticAIIntegration(), ], )

• MCP Python SDK (#4964) by @constantinius

  Enable the new Python MCP integration with the code snippet below:

  importsentry_sdkfromsentry_sdk.integrations.mcpimportMCPIntegrationsentry_sdk.init( dsn="<your-dsn>", # Set traces_sample_rate to 1.0 to capture 100%# of transactions for tracing.traces_sample_rate=1.0, # Add data like inputs and responses;# see https://docs.sentry.io/platforms/python/data-management/data-collected/ for more infosend_default_pii=True, integrations=[ MCPIntegration(), ], )

• fix(strawberry): Remove autodetection, always use sync extension (#4984) by @sentrivana

  Previously, StrawberryIntegration would try to guess whether it should install the sync or async version of itself. This auto-detection was very brittle and could lead to us auto-enabling async code in a sync context. With this change, StrawberryIntegration remains an auto-enabling integration, but it'll enable the sync version by default. If you want to enable the async version, pass the option explicitly:

  sentry_sdk.init( # ...integrations=[ StrawberryIntegration( async_execution=True ), ], )

• fix(google-genai): Set agent name (#5038) by @constantinius

• fix(integrations): hooking into error tracing function to find out if an execute tool span should be set to error (#4986) by @constantinius

• fix(django): Improve logic for classifying cache hits and misses (#5029) by @alexander-alderman-webb

• chore(metrics): Rename _metrics to metrics (#5035) by @alexander-alderman-webb

• fix(tracemetrics): Bump metric buffer size to 1k (#5031) by @k-fish

• fix startlette deprecation warning (#5034) by @DeoLeung

• build(deps): bump actions/upload-artifact from 4 to 5 (#5032) by @dependabot

• fix(ai): truncate messages for google genai (#4992) by @shellmayr

• fix(ai): add message truncation to litellm (#4973) by @shellmayr

• feat(langchain): Support v1 (#4874) by @sentrivana

• ci: Run common test suite on Python 3.14t (#4969) by @alexander-alderman-webb

• feat: Officially support 3.14 & run integration tests on 3.14 (#4974) by @sentrivana

• Make logger template format safer to missing kwargs (#4981) by @sl0thentr0py

• tests(huggingface): Support 1.0.0rc7 (#4979) by @alexander-alderman-webb

• feat: Enable HTTP request code origin by default (#4967) by @alexander-alderman-webb

• ci: Run common test suite on Python 3.14 (#4896) by @sentrivana

We are discontinuing development on 3.0. Please see
https://github.com/getsentry/sentry-python/discussions/4955 for more information.

3.0.0a7 is a maintenance release that adds a warning that there won't be
a stable 3.0 release. If you are on a 3.0 alpha release, please switch back
to 2.x to get the newest features and fixes.

Various fixes & improvements

• fix(gcp): Inject scopes in TimeoutThread exception with GCP (#4959) by @alexander-alderman-webb
• fix(aws): Inject scopes in TimeoutThread exception with AWS lambda (#4914) by @alexander-alderman-webb
• fix(ai): add message trunction to anthropic (#4953) by @shellmayr
• fix(ai): add message truncation to langgraph (#4954) by @shellmayr
• fix: Default breadcrumbs value for events without breadcrumbs (#4952) by @alexander-alderman-webb
• fix(ai): add message truncation in langchain (#4950) by @shellmayr
• fix(ai): correct size calculation, rename internal property for message truncation & add test (#4949) by @shellmayr
• fix(ai): introduce message truncation for openai (#4946) by @shellmayr
• fix(openai): Use non-deprecated Pydantic method to extract response text (#4942) by @JasonLovesDoggo
• ci: 🤖 Update test matrix with new releases (10/16) (#4945) by @github-actions
• Handle ValueError in scope resets (#4928) by @sl0thentr0py
• fix(litellm): Classify embeddings correctly (#4918) by @alexander-alderman-webb
• Generalize NOT_GIVEN check with omit for openai (#4926) by @sl0thentr0py
• ⚡️ Speed up function _get_db_span_description (#4924) by @misrasaurabh1