feat(api): update API spec from langfuse/langfuse b3283dc

[langfuse-bot]

Greptile Summary

The PR regenerates portions of the typed API client from an updated upstream specification.

• Adds session ID filtering to synchronous and asynchronous observation-list clients.
• Documents semantic-root metrics and evaluation-rule filtering.
• Clarifies comment-author membership and SCIM password behavior.
• Updates score endpoint documentation, including removal of existing v2 deprecation guidance.

Confidence Score: 4/5

The PR appears safe to merge, but its generated score-client documentation should retain the v2 deprecation and Langfuse v4 migration guidance.

The new observation filter is wired consistently and the boolean evaluation-filter model supports the documented shape; the remaining issue is non-blocking documentation that hides a known runtime incompatibility of unchanged v2 score routes.

Files Needing Attention: langfuse/api/scores/client.py and langfuse/api/scores/raw_client.py

Prompt To Fix All With AI

### Issue 1
langfuse/api/scores/client.py:164
**Removed v2 deprecation guidance**

These methods still call the v2 score endpoints, but this change removes the only visible warning that they are unavailable on Langfuse v4 and the guidance to use the v3 scores API, leaving affected users without the migration path when those calls fail.

---

For each issue above, determine whether it is valid and should be fixed. If so, fix it directly.

_{Reviews (1): Last reviewed commit: "feat(api): update API spec from langfuse..." | Re-trigger Greptile}

Context used:

• Knowledge Base — API Client (langfuse/api/)

[github-actions]

@claude review

[claude]

LGTM — this is an auto-generated API client sync from the Fern spec (langfuse/langfuse b3283dc), consisting entirely of docstring updates and new optional parameters passed through unchanged.

What was reviewed: the isRootObservation/sessionId params added to observations.get_many (both sync/async, client and raw_client) are correctly threaded through to the request params dict with proper camelCase aliasing; the new boolean filter type/docs additions to evaluation_rule_filter.py are documentation-only (the EvaluationRuleFilter_Boolean model already existed and is unchanged); and the scores.py docstring removals (dropping deprecation notices) don't change any runtime behavior.

Extended reasoning...

Overview

This PR is a generated update to the Langfuse Python API client, mirroring an upstream OpenAPI/Fern spec change in langfuse/langfuse (commit b3283dc). All 12 changed files are auto-generated client/raw_client/type files under langfuse/api/. The changes fall into four categories: (1) docstring clarifications for author_user_id (comments) and password (SCIM create_user) parameters, (2) new documentation for an isRootObservation v2-metrics dimension and a boolean filter type in evaluation rules, (3) a new optional session_id parameter added to observations.get_many and threaded through to the request params, and (4) removal of stale 'Deprecated' docstring banners from scores.get_many/get_by_id.

Security risks

None. There is no new authentication, authorization, or data-handling logic — this is a typed HTTP client whose changes are limited to docstrings, an added pass-through query parameter, and Pydantic model documentation. No user input is parsed or executed differently.

Level of scrutiny

Low. This is machine-generated boilerplate (noted by the 'auto-generated by Fern' header in every file) mirroring a spec change in the upstream API. The correct scrutiny here is verifying the generated diff is internally consistent (new params threaded through correctly, no accidental behavior changes) rather than deep logic review, since there is no hand-written business logic to reason about.

Other factors

I verified the new session_id parameter is consistently added to both sync/async get_many methods in client.py and raw_client.py, and correctly mapped to the sessionId query param. The scores.py deprecation-notice removal is docs-only and matches a plausible upstream reversal (endpoints previously marked deprecated for v4 are apparently still supported). No tests were added/changed, which is expected for this class of generated diff.