feat(api): update API spec from langfuse/langfuse 79d11de

[langfuse-bot]

No description provided.

[greptile-apps]

Too many files changed for review (126 files, 100 file limit).

Bypass the limit by tagging @greptile-apps to review.

[github-actions]

@claude review

[claude]

Looks good — mechanical, auto-generated Fern API client regeneration. I checked the new evaluator/evaluation-rule client wiring in langfuse/api/__init__.py and langfuse/api/client.py: every new type has matching entries in TYPE_CHECKING imports, _dynamic_imports, and __all__, and the new lazy-loaded evaluators/evaluation_rules client properties follow the exact same pattern as existing resources like datasets. Spot-checked evaluators/raw_client.py and the new package __init__.py files for consistent imports/exports; no hand-written logic, tests, or docs are touched.

Extended reasoning...

Overview

This PR is a bot-generated ("langfuse-bot") regeneration of the Fern-based Python API client, adding a new Evaluators/Evaluation Rules API surface plus supporting types (evaluation_commons, evaluation_errors) and unstable-preview mirrors. All 126 changed files live under langfuse/api/, following the exact same generated-code conventions (raw_client.py/client.py pairs, init.py with TYPE_CHECKING + _dynamic_imports + all pattern) used throughout the rest of the SDK.

Security risks

None identified. There is no new authentication, authorization, or crypto logic — these are pydantic model definitions and thin HTTP client wrappers that delegate to the existing SyncClientWrapper/AsyncClientWrapper infrastructure, unchanged in this diff. No user input handling or injection-relevant code paths are introduced.

Level of scrutiny

Low-to-moderate is appropriate: this is generated code mirroring an external OpenAPI spec, with a very mechanical, repetitive structure. I focused verification on internal consistency of the generated wiring (client.py property additions, init.py export/import maps) rather than on business logic, since there is no hand-authored logic here to scrutinize. I confirmed the new top-level client properties, dynamic import maps, and all lists are all mutually consistent and match the established pattern.

Other factors

No tests are included, but this matches the existing convention for this generated directory (no existing tests cover langfuse/api/ generated types either). The bug-hunting system reported no findings, and my own spot checks of the wiring code turned up no inconsistencies.