Skip to main content
Glama
deeneskoruzh
globalpocket

mcp-reranker

by globalpocket
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Remote

mcp-reranker

Ein generischer Model Context Protocol (MCP) Server, der Dokumenten-Reranking-Funktionen mittels sentence-transformers bereitstellt.

Dieser Server ist als eigenständiges Werkzeug konzipiert, das von jedem MCP-kompatiblen Client (wie Roo Code, Claude Desktop oder benutzerdefinierten Agenten) verwendet werden kann, um die Präzision von RAG (Retrieval-Augmented Generation) zu verbessern oder Agenten dabei zu helfen, bessere Entscheidungen zu treffen, indem die Relevanz zwischen einer Abfrage und mehreren Kandidaten bewertet wird.

Funktionen

  • Cross-Encoder Reranking: Nutzt das CrossEncoder-Modell von sentence-transformers für eine hochpräzise Relevanzbewertung.

  • Projektunabhängig: Vollständig unabhängig von spezifischer Anwendungslogik.

  • Anpassbare Modelle: Unterstützt verschiedene HuggingFace-Modelle. Sie können das Standardmodell über Umgebungsvariablen konfigurieren (Standard ist BAAI/bge-reranker-v2-m3).

  • JSON-Ausgabe: Gibt sortierte Ergebnisse in einem strukturierten JSON-Format zurück.

Related MCP server: ragi

Tools

rerank_documents

Berechnet Relevanz-Scores für eine Liste von Dokumenten im Vergleich zu einer gegebenen Abfrage und gibt diese sortiert nach Score zurück.

Argumente:

  • query (string): Die Suchanfrage oder die Kernabsicht, gegen die verglichen werden soll.

  • documents (array of strings): Eine Liste von Dokumentbeschreibungen oder Texten, die bewertet werden sollen.

  • model_name (string, optional): Die HuggingFace-Modellkennung. Standardmäßig wird die Umgebungsvariable RERANKER_MODEL_NAME oder "BAAI/bge-reranker-v2-m3" verwendet.

Antwortbeispiel: Ein JSON-formatierter String:

[
  { "document": "The most relevant document text.", "score": 0.985 },
  { "document": "A partially relevant text.", "score": 0.452 },
  { "document": "Completely irrelevant text.", "score": 0.012 }
]

Installation & Verwendung

Ausführung mit uvx

Fügen Sie Folgendes zu Ihrer MCP-Konfiguration hinzu (z. B. brownie_core_mcp_config.json). Sie können das verwendete Modell anpassen, indem Sie die Umgebungsvariable RERANKER_MODEL_NAME setzen.

{
  "mcpServers": {
    "mcp-reranker": {
      "command": "uvx",
      "args": [
        "--from",
        "git+[https://github.com/globalpocket/mcp-reranker.git](https://github.com/globalpocket/mcp-reranker.git)",
        "mcp-reranker"
      ],
      "env": {
        "RERANKER_MODEL_NAME": "BAAI/bge-reranker-v2-m3"
      }
    }
  }
}

Entwicklung

Voraussetzungen

  • Python 3.10+

  • uv

Einrichtung

git clone [https://github.com/globalpocket/mcp-reranker.git](https://github.com/globalpocket/mcp-reranker.git)
cd mcp-reranker
uv sync --extra dev

Tests ausführen

uv run pytest

Lizenz

MIT-Lizenz

Available Tools

1 tool
rerank_documentsA

Reranks a list of documents based on their relevance to a given query.

Args: query: The core intent or query string to compare against. documents: A list of document descriptions/strings to be ranked. model_name: The HuggingFace model name for the CrossEncoder. Default is from env RERANKER_MODEL_NAME or 'BAAI/bge-reranker-v2-m3'.

Returns: A JSON string representing a sorted list of dictionaries containing 'document' and 'score'.

ParametersJSON Schema
NameRequiredDescriptionDefault
queryYes
documentsYes
model_nameNo

Output Schema

ParametersJSON Schema
NameRequiredDescription
resultYes

TDQS

A4.1/5.0
Behavior3/5

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

With no annotations, the description must disclose behavioral traits. It does explain that it uses a CrossEncoder, mentions the model default, and describes the return format. However, it does not mention potential side effects like model downloads, network requirements, or that scores are sorted in descending order (only says 'sorted'). These are meaningful behavioral details that are missing.

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

Conciseness5/5

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

The description is well-structured and efficient: a one-sentence summary, clear Args, and Returns. No wasted words, and the core purpose is front-loaded. Every sentence serves a purpose.

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?

The description covers the essential inputs, behavior, and output format for a simple reranking tool. However, it omits the sort order (ascending/descending) and potential edge cases like empty document lists. Since an output schema exists, it is mostly complete but has minor gaps.

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

Parameters5/5

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

The input schema has no descriptions (0% coverage), but the description's Args section fully compensates by explaining each parameter, including the optional model_name and its default from env or 'BAAI/bge-reranker-v2-m3'. This adds complete semantic meaning beyond the schema.

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

Purpose5/5

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

The description clearly states the tool's function: 'Reranks a list of documents based on their relevance to a given query.' The verb 'reranks' is specific and the resource is clearly identified. Though no sibling tools are listed, the purpose is unambiguous and distinct.

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

Usage Guidelines3/5

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

The description implies when to use the tool (when you need to rerank documents by relevance) but does not explicitly state scenarios or provide exclusions. Since there are no sibling tools, explicit alternatives are not needed, but there is no guidance on when to prefer this tool over other potential approaches.

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

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 1 tool updatev0.1.2
    • First observedrerank_documents
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A4.2/5.0

Scored across 1 tool

Disambiguation5/5

Only one tool exists, so there is no possibility of confusion or overlap. The tool's purpose is clearly defined and distinct.

Naming Consistency5/5

The single tool name 'rerank_documents' follows a clear verb_noun convention and is descriptive. With only one tool, consistency is inherently maintained.

Tool Count3/5

The server has only one tool, which is at the low end of acceptable scope. While it fulfills its core reranking purpose, the count feels thin for a general utility server.

Completeness5/5

The tool fully covers the domain of reranking documents against a query. No additional operations are necessary for the stated purpose, making the surface complete.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • mcp-zeroentropy

    zeroentropy-ai
    -
    license
    Not graded
    quality
    Not graded
    maintenance
    MCP Server for ZeroEntropy collections, top documents and rerankers
    -

  • ragi

    SusuTawar
    A
    license
    A
    quality
    D
    maintenance
    Local-first RAG indexing and semantic search MCP server. Enables document retrieval and context-aware queries using local embedding models.
    3
    13 npm
    MIT

  • PDF RAG MCP Server

    abdellahiheiballa
    A
    license
    Not graded
    quality
    D
    maintenance
    A Model Context Protocol (MCP) server that provides powerful RAG (Retrieval-Augmented Generation) capabilities for PDF documents. This server uses ChromaDB for vector storage, sentence-transformers for embeddings, and semantic chunking for intelligent text segmentation.
    MIT