Bump the pip-dependencies group across 1 directory with 9 updates

[dependabot]

Updates the requirements on django, google-auth, sentry-sdk, resend, lxml, pillow, boto3, inspect-ai and datamodel-code-generator to permit the latest version.
Updates django to 6.1.1

Commits

• 249b13d [6.1.x] Bumped version for 6.1.1 release.
• 5f26fa8 [6.1.x] Added release date for 6.1.1.
• fdcf78a [6.1.x] Added remaining community package mentions to the documentation.
• 7241568 [6.1.x] Fixed #37222 -- Fixed QuerySet.distinct() crash on duplicated selecti...
• ef3fc80 [6.1.x] Fixed #37312, Refs #36605 -- Fixed annotation preservation and key se...
• b09cb6b [6.1.x] Clarified scope of object-level admin view permissions.
• a6d3aa2 [6.1.x] Corrected heading hierarchy in the admin actions documentation.
• 9031d41 [6.1.x] Clarified object-level permission checks in admin actions.
• fc805c6 [6.1.x] Fixed #37311 -- Prevented consumption of rhs iterators in annotation ...
• 4b0185a [6.1.x] Fixed #37300 -- Preserved parent instance hints on custom Prefetch qu...
• Additional commits viewable in compare view

Updates google-auth to 2.57.0

Release notes

Sourced from google-auth's releases.

google-auth: v2.57.0

2.57.0 (2026-08-21)

Features

• auth: add deprecation warning for grpcio < 1.83.0 (PQC support) (#18070) (68bdaba)

Bug Fixes

• auth: parse hostname for mTLS and PSC endpoint certificate rotat… (#18153) (b642373)
• auth: prevent TypeError and support home-dir cert fallback for X… (#18016) (b9a1379)

Documentation

• handwritten: centralize CONTRIBUTING.rst pointers (#17642) (23b9499)

Changelog

Sourced from google-auth's changelog.

Changelog

PyPI History

3.15.0 (2026-06-02)

Features

• regenerate google-cloud-d packages (#17092) (d49a2b9412b4ee9105125db36104a3e18344e0ad)

3.14.0 (2026-04-02)

Features

• Update comments for proto fields (c36aa6169035fdb14bf84b0e56e4f81291600b5a)
• Added a fields for image and table annotation output (c36aa6169035fdb14bf84b0e56e4f81291600b5a)

3.13.0 (2026-03-26)

Features

• Allow Protobuf 7.x (1eb7c268482e55971966e284dac2cbeb903adcbb)

Bug Fixes

• Require Python 3.9 (1eb7c268482e55971966e284dac2cbeb903adcbb)
• Require google-api-core >= 2.11.0 (1eb7c268482e55971966e284dac2cbeb903adcbb)
• Improve type checking (1eb7c268482e55971966e284dac2cbeb903adcbb)

3.12.0 (2026-03-23)

Features

• Add a field for upgrading previous processor version when fine tuning (ee7dd7dc168ecf943440cef38efff348e8eff095)

3.11.0 (2026-03-05)

Features

• Added a field for enabling image and table annotation for layout parser processor (6120fb0db084f5e8d1502fb1bb37594ac9976f6e)

3.10.0 (2026-02-12)

... (truncated)

Commits

• 3e29682 chore: release main (#18013)
• b4d9717 feat: check python and dependency versions in spanner packages (#18177)
• 66794a0 chore(bigtable): Rerouted legacy client row filters to data client row filter...
• b42122e Revert "fix(auth): parse hostname for mTLS and PSC endpoint certifica… (#18189)
• e20796a fix(google-auth-oauthlib): prevent port re-use on windows (#18166)
• b642373 fix(auth): parse hostname for mTLS and PSC endpoint certificate rotat… (#18153)
• 5bc3899 feat(spanner): add DataBoost and auto_partition_mode support to DBAPI driver ...
• 2b056ab docs(django-spanner, common-protos): centralize CONTRIBUTING.rst pointers (#1...
• 91d5e2a chore(bigtable): added warning to MutationsBatcher (#18176)
• 23b9499 docs(handwritten): centralize CONTRIBUTING.rst pointers (#17642)
• Additional commits viewable in compare view

Updates sentry-sdk to 2.68.1

Release notes

Sourced from sentry-sdk's releases.

2.68.1

enable_logs

• (logs) Don't stop sending auto-collected logs when enable_logs=True by @​sentrivana in #7237

  If you have enable_logs set to True, our logging integrations for the standard library logging module as well as Loguru will auto-collect logs and send them to Sentry as Sentry logs by default, preserving old behavior. Turning automatic collection off for a specific integration can still be achieved using the capture_sentry_logs integration option.

  import sentry_sdk
  from sentry_sdk.integrations.logging import LoggingIntegration
  from sentry_sdk.integrations.loguru import LoguruIntegration
  sentry_sdk.init(
  enable_logs=True,
  integrations=[
  LoggingIntegration(capture_sentry_logs=True),
  LoguruIntegration(capture_sentry_logs=False),
  ],
  )

  Please note that the enable_logs option is deprecated and will be removed in the next major release. The sentry_sdk.logger.X API now works regardless of it, and auto-collection can be opted into via the capture_sentry_logs integration-level options, which are False by default, unless you have enable_logs=True. We've added this compatibility layer to make the transition to a enable_logs-free world easier.

Bug Fixes 🐛

• (django) Add failed_request_status_codes by @​mgaligniana in #7140
• (anthropic) Gate gen_ai.response.tool_calls on outputs, not inputs by @​ericapisani in #7207
• (google_genai) Gate streaming gen_ai.response.tool_calls on outputs, not inputs by @​ericapisani in #7210
• (langchain) Use gen_ai.tool.definitions attribute when data collection is enabled by @​ericapisani in #7204
• (langgraph,huggingface_hub) Gate gen_ai.response.tool_calls on outputs, not inputs by @​ericapisani in #7208
• (openai) Gate gen_ai.response.tool_calls on outputs, not inputs by @​ericapisani in #7205
• (openai_agents) Gate gen_ai.response.tool_calls on outputs, not inputs by @​ericapisani in #7209

Changelog

Sourced from sentry-sdk's changelog.

2.68.1

enable_logs

• (logs) Don't stop sending auto-collected logs when enable_logs=True by @​sentrivana in #7237

  If you have enable_logs set to True, our logging integrations for the standard library logging module as well as Loguru will auto-collect logs and send them to Sentry as Sentry logs by default, preserving old behavior. Turning automatic collection off for a specific integration can still be achieved using the capture_sentry_logs integration option.

  import sentry_sdk
  from sentry_sdk.integrations.logging import LoggingIntegration
  from sentry_sdk.integrations.loguru import LoguruIntegration
  sentry_sdk.init(
  enable_logs=True,
  integrations=[
  LoggingIntegration(capture_sentry_logs=True),
  LoguruIntegration(capture_sentry_logs=False),
  ],
  )

  Please note that the enable_logs option is deprecated and will be removed in the next major release. The sentry_sdk.logger.X API now works regardless of it, and auto-collection can be opted into via the capture_sentry_logs integration-level options, which are False by default, unless you have enable_logs=True. We've added this compatibility layer to make the transition to a enable_logs-free world easier.

Bug Fixes 🐛

• (django) Add failed_request_status_codes by @​mgaligniana in #7140
• (anthropic) Gate gen_ai.response.tool_calls on outputs, not inputs by @​ericapisani in #7207
• (google_genai) Gate streaming gen_ai.response.tool_calls on outputs, not inputs by @​ericapisani in #7210
• (langchain) Use gen_ai.tool.definitions attribute when data collection is enabled by @​ericapisani in #7204
• (langgraph,huggingface_hub) Gate gen_ai.response.tool_calls on outputs, not inputs by @​ericapisani in #7208
• (openai) Gate gen_ai.response.tool_calls on outputs, not inputs by @​ericapisani in #7205
• (openai_agents) Gate gen_ai.response.tool_calls on outputs, not inputs by @​ericapisani in #7209

2.68.0

Important

• We're making enable_logs and enable_metrics no-op with this release (#7177), and they'll be dropped in the next major.

  Previously, enable_logs also controlled automatic logs collection from the logging and Loguru integrations. These integrations now get an integration-level capture_sentry_logs boolean option to allow for more control over the auto-collection. These options are False by default, i.e., nothing is auto-collected without your explicit opt-in.

  Action Needed

  If you had enable_logs set to True:

  • If you were using the sentry_sdk.logger.X API, no action necessary, the API will just work.
  • If you were auto-collecting logs from either LoggingIntegration or LoguruIntegration, the auto-collection will be turned off in this release. You can switch auto-collection on explicitly with:

... (truncated)

Commits

• 98f7d91 Update changelog
• 8f886ff release: 2.68.1
• df4ec02 fix(logs): Don't stop sending auto-collected logs when enable_logs=True (#7...
• e320318 test(pyramid): Add tests for data_collection gating of request body (#7218)
• 7b8a385 test(bottle): Add tests for data_collection gating of request body (#7217)
• cbcadc9 test(sanic): Add tests for data_collection gating of request body (#7216)
• e42c798 feat(tornado): Gate request body collection on data_collection option (#7215)
• d05bb60 test(django): Add tests for data_collection gating of request body (#7213)
• bcce67b feat(wsgi): Gate request body collection on data_collection option (#7212)
• 41220c4 fix(google_genai): Gate streaming gen_ai.response.tool_calls on outputs, no...
• Additional commits viewable in compare view

Updates resend to 2.42.0

Commits

• 09fa720 chore: bump version to 2.42.0 (#271)
• 45680b3 feat(webhooks): add event listing endpoints (#268)
• dc7632f chore: bump version to 2.41.0 (#266)
• 1c56643 feat: add update/update_async for renaming a segment (#270)
• 4c52e84 fix(api-keys): rename the update params key to id (#269)
• e3af5b9 chore: bump version to 2.40.1 (#267)
• fb80f87 fix: allow null contact first_name and last_name (#255)
• 36dcce6 fix(contacts): properties response shape (#254)
• a6441cc chore: bump version to 2.40.0 (#265)
• 533fd9c fix: take the broadcast id as a positional argument in recipients (#264)
• Additional commits viewable in compare view

Updates lxml to 6.1.3

Changelog

Sourced from lxml's changelog.

6.1.3 (2026-09-02)

Bugs fixed

• LP#2165901: External parameter entity parsing was allowed by default (with resolve_entities="internal"). Issue found by Tomer Fichman.

6.1.2 (2026-08-18)

• GH#526: Some build files were missing in the sdist. Patch by Nicola Soranzo.

• Some minor corrections for error handling cases.

Other changes

• Built with Cython 3.2.9.

6.1.1 (2026-05-18)

Bugs fixed

• The known link attributes in lxml.html.defs.link_attrs were missing xlink:href, which can be used for URL bypass attacks in embedded SVG/MathML/etc. content. GHSA-4jhm-jv67-739f

• The Linux wheels use a patched libxslt 1.1.43, fixing CVE-2025-7424 and CVE-2025-11731.

• The Windows wheels use libxslt 1.1.45, fixing CVE-2025-7424 and CVE-2025-11731.

6.1.0 (2026-04-17)

This release fixes a possible external entity injection (XXE) vulnerability in iterparse() and the ETCompatXMLParser.

Features added

• GH#486: The HTML ARIA accessibility attributes were added to the set of safe attributes

... (truncated)

Commits

• 3c1a4c7 Prepare release of 6.1.3.
• c1191fc Update changelog.
• 03ec312 Disable parameter entity parsing when internal-only entity parsing is requested.
• 11d03e9 Build: Prevent duplicate Py3.8 wheel builds.
• 9efc586 Build: Exclude musllinux-ARM from Py3.8 wheel building to prevent slow emulat...
• 9716fb1 Build: Include older PyPy versions.
• 0f3327d Build: Exclude Win-Aarch64 from wheel build.
• 6967c96 Build: Make all built wheels downloadable even if they don't pass the release...
• 061218d Build: Exclude Py3.15 i686 wheels from validation (because they are intention...
• ce9fe0d Build: Fix Py3.8 windows build.
• See full diff in compare view

Updates pillow to 12.3.0

Release notes

Sourced from pillow's releases.

12.3.0

https://pillow.readthedocs.io/en/stable/releasenotes/12.3.0.html

Removals

• Remove non-image ImageCms modes #9697 [@​radarhere]

Documentation

• Add release notes for SBOM and performance improvements #9747 [@​radarhere]
• Add security release notes #9741 [@​radarhere]
• Add release notes for Python 3.15 beta wheels #9696 [@​radarhere]
• ImageFont can also be used with ImageText #9597 [@​radarhere]
• Additional guidelines for security reports #9659 [@​wiredfool]
• Fixed typo #9636 [@​radarhere]
• Added CVEs to 12.2.0 release notes #9591 [@​radarhere]
• Revise development support information in README #9583 [@​aclark4life]
• Add INCIDENT_RESPONSE.md #9555 [@​aclark4life]
• Add STRIDE threat model to security docs #9562 [@​aclark4life]
• Add CVEs to 12.2.0 release notes #9556 [@​radarhere]
• Update README with revised security policy #9553 [@​radarhere]
• Update security policy #9552 [@​aclark4life]
• Update macOS tested Python versions #9534 [@​radarhere]

Dependencies

• Update dependency harfbuzz to v14.2.1 #9720 [@renovate[bot]]
• Update dependency mypy to v2 #9653 [@renovate[bot]]
• Update dependency cibuildwheel to v4 #9665 [@renovate[bot]]
• Update github-actions #9655 [@renovate[bot]]
• Update dependency libavif to v1.4.2 #9652 [@renovate[bot]]
• Update dependency lcms2 to v2.19.1 #9651 [@renovate[bot]]
• Update dependency check-jsonschema to v0.37.2 #9650 [@renovate[bot]]
• Update google/oss-fuzz digest to d872252 #9614 [@renovate[bot]]
• Update dependency lcms2 to v2.19 #9609 [@renovate[bot]]
• Update dependency libpng to v1.6.58 - autoclosed #9608 [@renovate[bot]]
• Update dependency harfbuzz to v14 #9610 [@renovate[bot]]
• Update dependency mypy to v1.20.2 #9599 [@renovate[bot]]
• Update github-actions #9611 [@renovate[bot]]
• Update dependency cibuildwheel to v3.4.1 #9607 [@renovate[bot]]
• Move dependency versions to single JSON and enable Renovate #9559 [@​hugovk]
• Updated raqm to 0.10.5 #9557 [@​radarhere]
• Update dependency cibuildwheel to v3.4.0 #9532 [@renovate[bot]]

Testing

• Remove matrix.os from benchmark #9735 [@​radarhere]
• Remove references to libavif patch #9734 [@​radarhere]
• Add benchmark tests #9654 [@​akx]
• Use reshape() instead of setting NumPy array shape directly #9728 [@​radarhere]

... (truncated)

Changelog

Sourced from pillow's changelog.

Changelog (Pillow)

11.1.0 and newer

See GitHub Releases:

• https://github.com/python-pillow/Pillow/releases

11.0.0 (2024-10-15)

• Update licence to MIT-CMU #8460 [hugovk]

• Conditionally define ImageCms type hint to avoid requiring core #8197 [radarhere]

• Support writing LONG8 offsets in AppendingTiffWriter #8417 [radarhere]

• Use ImageFile.MAXBLOCK when saving TIFF images #8461 [radarhere]

• Do not close provided file handles with libtiff when saving #8458 [radarhere]

• Support ImageFilter.BuiltinFilter for I;16* images #8438 [radarhere]

• Use ImagingCore.ptr instead of ImagingCore.id #8341 [homm, radarhere, hugovk]

• Updated EPS mode when opening images without transparency #8281 [Yay295, radarhere]

• Use transparency when combining P frames from APNGs #8443 [radarhere]

• Support all resampling filters when resizing I;16* images #8422 [radarhere]

• Free memory on early return #8413 [radarhere]

• Cast int before potentially exceeding INT_MAX #8402 [radarhere]

... (truncated)

Commits

• bb1d8e8 12.3.0 version bump
• e63fc48 Add release notes for SBOM and performance improvements (#9747)
• 13b701b Add release notes for #9679
• 5564ca7 List methods
• a0920fd Speed up ImageChops operations (#9738)
• 07e9a6c Speed up Image.filter() (#9736)
• a94578c Speed up Image.getchannel(), Image.merge(), Image.putalpha() and `Image...
• 53e02c4 Speed up Image.fill(), Image.linear_gradient() and `Image.radial_gradient...
• af03747 Speed up Image.resample() (#9739)
• 5c9ca56 Speed up alpha_composite, matrix, negative, quantize (#9740)
• Additional commits viewable in compare view

Updates boto3 to 1.43.88

Commits

• bb429b1 Merge branch 'release-1.43.88'
• e373d8d Bumping version to 1.43.88
• 6ce83de Add changelog entries from botocore
• d5740c2 Merge branch 'release-1.43.87'
• bd30961 Merge branch 'release-1.43.87' into develop
• 316b72c Bumping version to 1.43.87
• 17409b4 Add changelog entries from botocore
• 81ae047 Merge branch 'release-1.43.86'
• 0e52e63 Merge branch 'release-1.43.86' into develop
• 938cd05 Bumping version to 1.43.86
• Additional commits viewable in compare view

Updates inspect-ai to 0.3.262

Changelog

Sourced from inspect-ai's changelog.

0.3.262 (02 September 2026)

• Scorer (breaking): Model-graded scorers with a panel of graders now require a strict majority: samples with no majority grade (even-split ties, three-way splits, or ties after a grader returns no parseable grade) come back unscored and leave the metric denominator, rather than the most common grade winning with ties broken by grader order. Pass reducer="mode" to model_graded_qa()/model_graded_fact() to restore the previous behavior. (#4721)
• Scorer: pattern() now falls back to the full regex match when the pattern contains no explicit capture groups (previously such patterns always scored INCORRECT). (#4828)
• Bugfix: Bump the fsspec upper bound from <=2025.9.0 to <=2026.6.0 to align with the current huggingface/datasets cap. (#4761)
• Concurrency: max_sandboxes and max_subprocesses now default off the processors the eval may actually use, so an eval in a CPU-limited container no longer oversubscribes its quota.
• Tools: The grep tool now supports extended regex via a new extended_regexp option (patterns remain basic regex by default).
• Tools: Tool call arguments nested deeper than 100 levels are now rejected with a parse error echoed back to the model (previously depths up to ~254 executed and deeper crashed eval logging, aborting the whole run).
• Agent Bridge: Bridged Anthropic requests now preserve system block boundaries, so instruction blocks are no longer silently discarded by the API.
• Anthropic: A single assistant turn that interleaves thinking with client tool calls now replays in its original order, so subsequent requests no longer fail with "thinking ... blocks in the latest assistant message cannot be modified".
• Anthropic: Support for Claude Fable 5.1 and Mythos 5.1 — forced tool choice degrades to auto on both, and history edits no longer fail Fable 5.1's replayed thinking blocks (first-party API).
• Google: Added model info for Gemini 3.8 Flash and Gemini 3.7 Flash; reasoning_effort="minimal" now maps to low (with a warning) on models that reject minimal thinking rather than failing the request.
• OpenAI: Chat Completions usage now preserves prompt-cache read and write tokens for accurate cache-aware costing.
• Scoring: model_graded_qa/model_graded_fact no longer score a malformed multi-character verdict such as GRADE: CI as correct; such verdicts now leave the sample unscored with grade_parse_failure recorded.
• OpenAI: Fixed background responses failing on transient connection and timeout errors.
• Eval Log: local_path() now percent-decodes file:// URLs, restoring the to_uri() round trip for paths with spaces or other encoded characters (affects every consumer of file URLs: log recorders, control-channel state, checkpoint restore, approval policy files). (#5025)
• Bugfix: Sample files values that are empty strings now create empty files in the sandbox rather than recursively copying the runner's working directory (or the dataset's directory) into it.
• Scoring: Support tuples and non-list sequences in Target. (#5039)
• Analysis: Fixed task_info() and log_viewer() raising ValueError when applied to an empty DataFrame. (#4826)
• Eval Set: eval_set() now defaults log_dir to INSPECT_LOG_DIR or ./logs, as eval() does, rather than requiring it.
• Eval Set: An external runner driving eval_set() can now override any argument that does not change task identity, rather than only five.
• Eval Set: An eval set driven by an external runner now honors the INSPECT_EVAL_* environment variables with the same meanings inspect eval-set gives them.
• Eval Set: Eval set selections now support specifying scanners.
• Control Channel: inspect ctl config --max-samples now works for tasks using adaptive connections — an integer pins sample concurrency, and clear resumes adaptive tracking.
• Eval logs: Header-only .eval log uploads to S3 now appear in inspect trace output.
• MCP: A shared ToolSource no longer serves one sample's tools to another; tools are now resolved fresh through the server on every call rather than cached. Sessions are scoped per async task rather than per sample, so a handoff or other subagent's turn (which runs in its own task) now gets its own connection to a local MCP server rather than reusing its parent's live session.
• Bugfix: Sustained rate limiting no longer pins the adaptive connection limit after a single reduction; it keeps adapting up and down as conditions change.
• Eval Log: Resolving a log's attachments no longer discards model call payloads, which previously became unreadable when attachments were resolved.
• Eval Log: Fixed trio evals crashing with ValueError: seek of closed file when writing a .eval log smaller than 8MB to S3.
• Eval Log: Writing a sample whose content cannot be serialized (e.g. score metadata nested beyond pydantic's depth limit) no longer aborts the eval at flush or finish; the sample is logged with the offending content removed and the removal recorded in its error.
• Eval Log: JSON values in logged events nested deeper than 240 levels are now truncated with a placeholder marker (previously depths up to ~254 were preserved and deeper crashed the eval).
• Inspect CTL: New inspect ctl sample score interim-scores a single running sample's work-so-far on demand (briefly held while scored; the sample keeps running).
• Score: inspect score now reports samples that errored or were stopped early, so a re-scored partial run is no longer displayed as if it were complete.
• Scorers: New cascade() scorer runs scorers in order and reports the first that settles a sample, so an expensive grader only runs on samples cheaper scorers left unsettled.
• Agent bridge: Anthropic responses now report thinking tokens in usage.output_tokens_details, so bridged clients can distinguish a reasoning response from a plain one.
• Agent bridge: A bridged request asking for Anthropic output_config.format, OpenAI text.verbosity, or any of Gemini's thinkingConfig.thinkingBudget, presencePenalty, frequencyPenalty, responseLogprobs and logprobs now gets it, instead of silently receiving the model default.
• Agent bridge: A bridged Gemini response now reports thoughtsTokenCount, so a client can see how many thinking tokens a request used.
• Agent bridge: An invalid generation parameter in a bridged request (e.g. stop_sequences: 5) is now rejected with a 400 instead of being recorded into an event that cannot be read back, which made the whole sample transcript unreadable.
• Agent bridge: A bridged Gemini response's candidatesTokenCount no longer double-counts thinking tokens alongside thoughtsTokenCount.
• Agent bridge: A bridged structured-output request whose JSON Schema uses keywords Inspect does not model (e.g. $ref) now warns, instead of silently constraining the model more weakly than asked.
• Agent bridge: An invalid response schema or mistyped request field in a bridged OpenAI, Anthropic, or Google request is now rejected with a 400 instead of escaping as a status-less internal error.
• Eval Log: Buffer manifest segment entries are now TypedDicts rather than pydantic models, cutting manifest parse time and GC pressure on the sync thread for runs with many segments.
• Eval Log: Buffer manifests are no longer written with indentation, which accounted for ~41% of their bytes on every log_shared sync.
• Agent bridge: Fixed a tool-approval bypass where a sandboxed agent could run approval-gated bridged_tools by calling the bridge service directly; unapproved host tool calls are now denied.
• Bugfix: Approval policies set on a Task are now recorded in the eval log.
• OpenAI: Responses API requests now omit the id key on synthesized message and reasoning items rather than sending an explicit null, which backends like vLLM reject.
• Hooks: Fixed hooks published by extension packages silently not loading when another hook was already registered at startup.
• Bugfix: Model calls no longer time out during long samples with realtime logging or bounded transcripts enabled.
• Control Channel: inspect ctl sample cancel --action cancel now works on samples that haven't started — cancelling a never-started sample before it runs and withdrawing (un-requeuing) a queued re-run so its prior outcome stands.
• Crash Recovery: inspect log recover, eval-retry, and eval_set gain --incomplete-action error (with an --incomplete-max guard) to resolve samples that were in progress at a crash as errors and finalize the eval as success instead of endlessly re-running hung samples.

... (truncated)

Commits

• f2fb4ad update changelog for release
• 0494dad Restore the 0.3.261 release section to CHANGELOG.md (#5205)
• 1de5f74 Support for Gemini 3.8 Flash and Gemini 3.7 Flash (reasoning_effort=minimal n...
• b30a4f5 Stop re-parsing the model-info database once per test; 2026-08-29 / 08-31 / 0...
• d2edd71 Fix four bugs in the suppressions PR comment job (#5151)
• 38401bb ci: assert the timeline fixture corpus is present in viewer-tests (#5197)
• d2a2a8f Resolve stalled samples at recovery: incomplete-action dispositions finalize ...
• e2a8e85 Support for Claude Fable 5.1 and Mythos 5.1 (#5176)
• 76f1aa7 Fix tool-approval bypass for host tools called through the sandbox agent brid...
• 507db1c Fall back to auto tool_choice on non-K3 Kimi models with thinking enabled (#5...
• Additional commits viewable in compare view

Updates datamodel-code-generator to 0.76.1

Release notes

Sourced from datamodel-code-generator's releases.

0.76.1

What's Changed

• Update CHANGELOG for 0.76.0 by @​dcg-generated-docs[bot] in koxudaxi/datamodel-code-generator#3812
• Update release benchmark data by @​dcg-generated-docs[bot] in koxudaxi/datamodel-code-generator#3813
• Fix payload runtime validation exclusions by @​koxudaxi in koxudaxi/datamodel-code-generator#3815
• Optimize collapsed root model replacement by @​koxudaxi in koxudaxi/datamodel-code-generator#3826
• Track collapsed root model references incrementally by @​koxudaxi in koxudaxi/datamodel-code-generator#3827
• Skip unnecessary built-in string normalization by @​koxudaxi in koxudaxi/datamodel-code-generator#3828
• Fix runtime validator output model ownership by @​koxudaxi in koxudaxi/datamodel-code-generator#3817
• Move parser source loading by @​koxudaxi in koxudaxi/datamodel-code-generator#3818
• Move field name policies to output models by @​koxudaxi in koxudaxi/datamodel-code-generator#3820
• Decouple parser run lifecycle by @​koxudaxi in koxudaxi/datamodel-code-generator#3821
• Move parser recursion signal by @​koxudaxi in koxudaxi/datamodel-code-generator#3822
• Move output template context to models by @​koxudaxi in koxudaxi/datamodel-code-generator#3823
• Isolate parser source context by @​koxudaxi in koxudaxi/datamodel-code-generator#3824
• Decouple input model transport by @​koxudaxi in koxudaxi/datamodel-code-generator#3825
• Enforce architecture boundaries by @​koxudaxi in koxudaxi/datamodel-code-generator#3832
• Allow null enum descriptions by @​koxudaxi in koxudaxi/datamodel-code-generator#3835
• Fix release draft analysis for stacked PRs by @​koxudaxi in koxudaxi/datamodel-code-generator#3836
• Fix paginated release draft reads by @​koxudaxi in koxudaxi/datamodel-code-generator#3837
• Fix release draft PR permissions by @​koxudaxi in koxudaxi/datamodel-code-generator#3847

Full Changelog: koxudaxi/datamodel-code-generator@0.76.0...0.76.1

Changelog

Sourced from datamodel-code-generator's changelog.

0.76.1 - 2026-09-02

What's Changed

• Update CHANGELOG for 0.76.0 by @​dcg-generated-docs[bot] in koxudaxi/datamodel-code-generator#3812
• Update release benchmark data by @​dcg-generated-docs[bot] in koxudaxi/datamodel-code-generator#3813
• Fix payload runtime validation exclusions by @​koxudaxi in koxudaxi/datamodel-code-generator#3815
• Optimize collapsed root model replacement by @​koxudaxi in koxudaxi/datamodel-code-generator#3826
• Track collapsed root model references incrementally by @​koxudaxi in koxudaxi/datamodel-code-generator#3827
• Skip unnecessary built-in string normalization by @​koxudaxi in koxudaxi/datamodel-code-generator#3828
• Fix runtime validator output model ownership by @​koxudaxi in koxudaxi/datamodel-code-generator#3817
• Move parser source loading by @​koxudaxi in koxudaxi/datamodel-code-generator#3818
• Move field name policies to output models by @​koxudaxi in koxudaxi/datamodel-code-generator#3820
• Decouple parser run lifecycle by @​koxudaxi in koxudaxi/datamodel-code-generator#3821
• Move parser recursion signal by @​koxudaxi in koxudaxi/datamodel-code-generator#3822
• Move output template context to models by @​koxudaxi in koxudaxi/datamodel-code-generator#3823
• Isolate parser source context by @​koxudaxi in koxudaxi/datamodel-code-generator#3824
• Decouple input model transport by @​koxudaxi in koxudaxi/datamodel-code-genera...

  Description has been truncated

[github-actions]

Warning

WIP limit exceeded. There are now 7 open pull requests — the project target is 5 or fewer.

Please close or merge an existing PR before continuing with this one.