Feature Request: Replace `get_entity` with `get_entities` for Batch Processing

[dmytroDragan]

Overview

Most MCP clients currently use get_entity in a loop, resulting in one API call per entity per execution. This pattern introduces unnecessary overhead and slows down processing.

Request

The replacement or supplementing get_entity with a get_entities method that supports batch retrieval of multiple entities in a single call.

Motivation

• Efficiency: Reduces the number of API calls, lowering latency.
• Performance: Speeds up client execution times, especially for large datasets.
• Resource Usage: Decreases load on both client and server.

[hsheth2]

I'm thinking about this one - you're right that it's a common pattern and reducing the round trips would make sense. The main worry I have is that getting a bunch of entities in bulk could consume a lot of tokens, which may not be desirable.

[dmytroDragan]

Thank you for feedback.

Token limit could indeed become an issue. I'm thinking on adding parameters similar to Datahub batch OpenAPI:

1. count, scroll id/offset and query/filter criteria - to control the number of objects
2. list of properties - to minimize the context only to required fields. (which would also require a possiblity to provide entity schemas in the first place by Datahub GraphQL API introspection or pdl-based aspect list )

it will provide more control, but also add more complexity.

Another point is that a batching becomes a part of new specification: https://modelcontextprotocol.io/specification/2025-03-26/changelog#major-changes