feat(api): update API spec from langfuse/langfuse 9ce4aa4

[langfuse-bot]

Greptile Summary

This is an auto-generated API spec sync that updates docstrings across the LLM connections and observations modules. No runtime logic is changed.

• Adds OpenAI adapter documentation to the config field across LlmConnection, UpsertLlmConnectionRequest, and all client wrappers, describing the optional {\"useResponsesApi\": <boolean>} config shape.
• Refines the matches operator description in observations clients from "token-based full-text search" to "indexed literal search with token-boundary pruning", and explicitly notes that % and _ are treated as literal characters (unlike SQL LIKE).

Confidence Score: 5/5

This PR is safe to merge — all changes are docstring updates generated from the upstream API spec with no runtime logic affected.

Every changed line is inside a docstring or inline comment. The OpenAI useResponsesApi config option is additive documentation, and the matches operator clarification corrects the description of existing search behavior without altering any code path.

No files require special attention. The minor formatting difference in llm_connection.py (uses a concrete true example) vs the other files (use placeholder) is cosmetic and generated from the upstream spec.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[UpsertLlmConnection / LlmConnection] --> B{adapter type?}
    B -->|Bedrock| C["config REQUIRED\n{region: aws-region}"]
    B -->|OpenAI| D["config OPTIONAL\n{useResponsesApi: boolean}\nRoutes via OpenAI Responses API"]
    B -->|VertexAI| E["config OPTIONAL\n{location: gcp-location}"]
    B -->|Other| F["config NOT SUPPORTED\nomit or set null"]

    G[Observations filter: matches operator] --> H{field?}
    H -->|input / output| I["Indexed literal search\ncase-insensitive\n% and _ are literals"]
    H -->|metadata stringObject| J["Indexed literal search\ncase-sensitive"]
    H -->|other| K["Not supported"]
    I --> L["Must be accompanied by\n= or matches on input/output"]

Loading

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

[github-actions]

@claude review

[claude]

Auto-generated API spec update — docstring-only changes.

Extended reasoning...

Overview

This PR is an auto-generated API spec update (Fern). It modifies only docstrings in 6 auto-generated files: clarifying the LLM connections config field to document a new OpenAI useResponsesApi option, and refining the description of the observations matches filter operator to describe its semantics (indexed literal search, literal %/_).

Security risks

None. There are no logic, signature, parameter, request payload, or wire-format changes — only prose updates inside Python docstrings of generated client code.

Level of scrutiny

Minimal. These files are clearly marked as auto-generated by Fern from the API definition, and the diff is purely textual documentation. No callers behavior changes, no new fields, no validation logic.

Other factors

The bug hunting system found no issues. The PR follows the established pattern of routine Fern-generated spec sync PRs (cf. recent commit fa4d81a with the same shape).