RushDB MCP Server

Official

Overview Schema Source Related Servers Discussions Admin

vectorSearch

Read-only

Search RushDB records by vector similarity on an indexed property, using a text query or precomputed vector, with optional filters to narrow and rank results.

Instructions

Perform direct vector similarity search over records whose propertyName has been indexed with createEmbeddingIndex. For managed indexes: provide a free-text query — RushDB embeds it and returns the most similar records ranked by similarity (__score). For external indexes: provide a queryVector (pre-computed number[]) instead of query text. A where filter narrows candidates before ranking. Requires an embedding index in "ready" status for the given label+propertyName.

Input Schema

TableJSON Schema

Name	Required	Description
`ef`	No	ANN gather breadth when the search runs on the ANN graph.
`mode`	No	How to rank candidates. "auto" (default): exact on small indexes, the ANN graph on large ones. "exact"/"ann" force a path ("ann" needs a cosine index).
`rrfK`	No	Hybrid only: the RRF smoothing constant (default 60).
`skip`	No	Number of results to skip for pagination (default 0).
`text`	No	Adds a keyword (BM25) leg, making it a hybrid search: both rankings run over the same filtered records and are fused. `query` defaults to the semantic query; `index` defaults to the one ready text index of the first label (see listTextIndexes).
`index`	No	Name of the embedding index to search (as listed by findEmbeddingIndexes or getSchema). When given, propertyName and labels default to the index's own.
`limit`	No	Maximum number of results to return (default 10).
`query`	No	Free-text query to embed and compare against stored vectors (managed indexes only). Mutually exclusive with queryVector.
`where`	No	Optional filter applied before scoring (activates prefilter mode).
`fusion`	No	Hybrid only: "rrf" (default) fuses ranks; "dbsf" fuses each leg's scores normalized by its own distribution.
`labels`	No	One or more record labels to scope the search. The first label is used to resolve the embedding index.
`weight`	No	Hybrid only (with `text`): the vector leg's share of the fused score, ≥ 0 (default 1).
`dimensions`	No	Disambiguates when multiple indexes share the same label+property. Inferred from queryVector length when queryVector is supplied.
`sourceType`	No	Select which index type to search. Required when both managed and external indexes exist for the same label+property.
`queryVector`	No	Pre-computed embedding vector to search with (external indexes). Mutually exclusive with query. Length must match index dimensions.
`propertyName`	No	Name of the indexed property to search against. Required unless index is given.
`prefetchLimit`	No	Hybrid only: hits the vector leg ranks before fusion, 1–10,000 (default 4 × (skip + limit)).
`scoreThreshold`	No	Keep only results scoring at least this much (scores depend on the embedding model).
`similarityFunction`	No	Disambiguates when multiple indexes share the same label+property+sourceType.

Output Schema

TableJSON Schema

Name	Required	Description	Default
`data`	Yes
`total`	Yes

Schema Changelog

Changes observed during successful MCP inspections.

First observedv3.0.1Oct 21, 2026

TDQS

A3.9/5.0

Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare readOnlyHint=true/openWorldHint=false, so safety is covered; the description adds real context beyond that — results are ranked by similarity with a __score, `where` triggers prefilter mode, hybrid fusion runs both legs over the same filtered records, and the tool fails without a ready index. Missing details on failure modes or performance characteristics.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Four sentences, front-loaded with the core action and scope before moving to the managed/external branching and the prerequisite. Dense but each sentence carries information; only the prerequisite could arguably be hoisted earlier.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 19-parameter read tool with an output schema, the description covers the two invocation paths, the index prerequisite, and the prefilter/hybrid behavior an agent needs to choose arguments. Since the output schema carries return values, the only real gap is the absence of guidance against the neighboring search tools.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the schema already documents all 19 parameters including the query/queryVector mutual exclusivity and the managed/external sourceType. The description restates the managed-vs-external choice and the prefilter role of `where`, adding modest framing value but no parameter meaning the schema lacks — baseline 3 applies.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource: 'Perform direct vector similarity search over records whose propertyName has been indexed'. The word 'direct' and the createEmbeddingIndex reference hint at how it differs from higher-level siblings like semanticSearch or smartSearch, but it never names those siblings, so the differentiation is inferential rather than explicit.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives clear conditional guidance: use free-text `query` for managed indexes, `queryVector` for external indexes, and states the prerequisite that an embedding index must be in 'ready' status for the label+propertyName. It does not name alternative tools (semanticSearch, smartSearch, textSearch) or say when NOT to use this, so it stops short of full routing guidance.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.