[langfuse-bot]

Generated from langfuse/langfuse main at 832e97a.

Confidence Score: 5/5

The PR appears safe to merge; no actionable issues were found.

Summary

Updates the generated API client from the upstream API definition.

• Python callers can list Topics rules and fetch one by ID.
• Python callers can query the latest Topics summary for each trace and facet.
• Python callers can query current topic assignments for traces and facets.
• Topics responses identify linked rules and describe current-map topics.
• Observation v2 explains when callers must keep fetching cursor pages.

No actionable issues were found. Review was based on source inspection; no runtime tests were run.

_{Reviews (1) · Last reviewed commit: "feat(api): update API spec from langfuse..." · Reviewed by Greptile}

[github-actions]

@claude review

[claude]

Claude Code Review

No review was started: this request came from a bot account. Manual reviews can only be requested by someone with write access to this repository. Ask a maintainer to comment @claude review, or have your automation post the comment from a user account with write access.

_{Tip: disable this comment in your organization's Code Review settings.}

[claude]

🔴 With this SDK, list_facets and get_facet raise a pydantic ValidationError against any Langfuse server that does not yet send ruleIds; on the base branch they parse fine. TopicFacet.rule_ids is declared required with pydantic.Field() and no default. parse_obj_as validates responses strictly, so a facet without that key fails. Fix: make rule_ids optional (typing.Optional[typing.List[str]] = None) or default it to an empty list. Do the same in the API spec, so self-hosted or older servers still parse.

Why this was flagged

This applies only when the server runs a Langfuse build older than spec 832e97a, which omits ruleIds from facets. Examples are self-hosted deployments and closed-beta instances not yet upgraded. TopicFacet.rule_ids at langfuse/api/unstable/topics/types/topic_facet.py:22-24 has no default. TopicFacetList.data is typing.List[TopicFacet]. RawTopicsClient.list_facets and get_facet call parse_obj_as(type_=TopicFacetList / TopicFacet), which runs pydantic.TypeAdapter(type_).validate_python (langfuse/api/core/pydantic_utilities.py:58-59). That raises ValidationError for the missing field. The surrounding try in raw_client.py catches only JSONDecodeError, so the error reaches the caller. On the base branch the same response parses, because rule_ids did not exist. Every other field added in this diff (Topic.rule_id, TopicFacetCounts.rule_id, ObservationsV2Meta.comment) is Optional with a default. extra="allow" tolerates unknown keys, not missing required ones.

Verification: This diff adds rule_ids with pydantic.Field() in langfuse/api/unstable/topics/types/topic_facet.py at lines 22-24, so the field is required. In raw_client.py, list_facets (lines 110-114) and get_facet (lines 289-293) parse with parse_obj_as, which raises a ValidationError on a missing required field. On the base commit the same response parsed fine.