Skip to main content
Glama
masseater

Gemini RAG MCP Server

by masseater
OverviewSchemaRelated ServersScoreDiscussions
TypeScript
Remote

Gemini RAG MCP Server

A Model Context Protocol (MCP) server that provides RAG (Retrieval-Augmented Generation) capabilities using Google's Gemini API File Search feature. This server enables AI applications to create knowledge bases and retrieve information from uploaded documents.

Features

  • ✅ File Search RAG: Create and manage knowledge bases using Gemini's File Search API

  • ✅ Document Upload: Upload files and text content to create searchable knowledge bases

  • ✅ Information Retrieval: Query knowledge bases to retrieve relevant information

  • ✅ Configurable Models: Choose Gemini models via environment variable

  • ✅ MCP Protocol: Full compatibility with Model Context Protocol

  • ✅ Type-Safe: Full TypeScript support with strict mode enabled

  • ✅ Dual Transport Support: stdio (default) and HTTP transports

  • ✅ Production-Ready: Logging, error handling, and configuration management

Related MCP server: MCP RAG Server

Prerequisites

  • Node.js >= 22.10.0

  • pnpm >= 10.19.0

  • Google API Key with Gemini API access

Installation

Using with Claude Desktop (Recommended)

Add the following to your Claude Desktop configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "gemini-rag-mcp": {
      "command": "npx",
      "args": ["-y", "@r_masseater/gemini-rag-mcp"],
      "env": {
        "GOOGLE_API_KEY": "your_google_api_key_here",
        "STORE_DISPLAY_NAME": "your_store_name"
      }
    }
  }
}

Required Environment Variables:

  • GOOGLE_API_KEY: Your Google API key with Gemini API access

  • STORE_DISPLAY_NAME: Display name for your vector store/knowledge base

Optional Environment Variables:

  • GEMINI_MODEL: Gemini model to use for queries (default: gemini-2.5-pro)

    • Options: gemini-2.5-pro, gemini-2.5-flash

After configuration, restart Claude Desktop to load the server.

Development

1. Clone the repository

git clone https://github.com/masseater/gemini-rag-mcp.git
cd gemini-rag-mcp

2. Install dependencies

pnpm install

3. Run in development mode

# stdio transport (default)
pnpm run dev

# HTTP transport (with hot reload)
pnpm run dev:http

Environment Variables

Required:

  • GOOGLE_API_KEY: Google API key with Gemini API access

  • STORE_DISPLAY_NAME: Display name for vector store/knowledge base

Optional:

  • GEMINI_MODEL: Gemini model for queries (default: gemini-2.5-pro)

  • LOG_LEVEL: Logging level (error|warn|info|debug, default: info)

  • DEBUG: Enable debug console output (true|false, default: false)

  • PORT: HTTP server port (default: 3000)

Available Tools

Once configured with Claude Desktop, the following tools are available:

  • upload_file: Upload document files to the knowledge base

  • upload_content: Upload text content directly to the knowledge base

  • query: Query the knowledge base using RAG

Resources

License

MIT License

Available Tools

3 tools
queryB

Query the FileSearchStore using RAG (Retrieval-Augmented Generation) to get answers based on uploaded documents. The AI will search through the documents and provide relevant answers with citations.

ParametersJSON Schema
NameRequiredDescriptionDefault
queryYesThe question or query to search for in the knowledge base

TDQS

B3.2/5.0
Behavior2/5

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

No annotations are provided, so the description carries full burden. It discloses the tool uses RAG for searching and provides answers with citations, which adds some behavioral context beyond the input schema. However, it lacks details on permissions, rate limits, error handling, or response format (beyond 'answers with citations'), which are critical for a query tool. The description doesn't contradict annotations (none exist).

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?

The description is two sentences, front-loaded with the core purpose and followed by additional detail on how it works. It avoids redundancy and wastes no words, though it could be slightly more structured (e.g., separating usage notes). Every sentence adds value: the first defines the tool, and the second explains the mechanism and output.

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

Completeness3/5

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

Given the tool's moderate complexity (RAG-based querying), no annotations, no output schema, and a simple input schema, the description is minimally adequate. It covers the purpose and basic behavior but lacks completeness in areas like output details, error cases, or integration with sibling tools. It meets the minimum viable threshold but has clear gaps for effective agent use.

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?

The input schema has 100% description coverage, with the single parameter 'query' documented as 'The question or query to search for in the knowledge base.' The description adds no additional parameter semantics beyond this, such as examples or constraints on query format. With high schema coverage, the baseline is 3, as the schema already provides adequate parameter information.

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?

The description clearly states the tool's purpose: 'Query the FileSearchStore using RAG (Retrieval-Augmented Generation) to get answers based on uploaded documents.' It specifies the verb ('query'), resource ('FileSearchStore'), and method ('RAG'), distinguishing it from sibling tools 'upload_content' and 'upload_file' which are for uploading rather than querying. However, it doesn't explicitly contrast with hypothetical alternative query methods, keeping it at 4 instead of 5.

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 usage context: 'to get answers based on uploaded documents,' suggesting this tool should be used after documents are uploaded via sibling tools. It doesn't provide explicit when-not-to-use guidance or name alternatives for different query types, but the implied dependency on uploaded content offers basic context. No misleading information is present.

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

upload_contentC

Upload text content to the FileSearchStore for RAG indexing. The content will be processed and made searchable.

ParametersJSON Schema
NameRequiredDescriptionDefault
contentYesText content to upload to the knowledge base
displayNameYesDisplay name for the content in the store
metadataNoCustom metadata as key-value pairs. Values can be strings or numbers. Maximum 20 entries per document. Example: {"category": "guide", "year": 2025}

TDQS

C2.9/5.0
Behavior2/5

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

With no annotations provided, the description carries full burden but offers minimal behavioral insight. It mentions processing and making content searchable but lacks details about permissions, rate limits, idempotency, error conditions, or what 'processed' entails. For a write operation with zero annotation coverage, this is inadequate.

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 extremely concise with two sentences that directly communicate the core functionality. Every word earns its place, and the information is front-loaded with no wasted verbiage.

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

Completeness2/5

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

For a write operation with no annotations and no output schema, the description is insufficient. It doesn't explain what happens after upload (e.g., success indicators, returned IDs, error responses), nor does it provide behavioral context needed for safe and effective use.

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%, providing good documentation for all parameters. The description adds no parameter-specific information beyond what's in the schema, so it meets the baseline of 3 where the schema does the heavy lifting.

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?

The description clearly states the action ('upload text content') and destination ('to the FileSearchStore for RAG indexing'), with a specific purpose ('made searchable'). It distinguishes from 'upload_file' by specifying text content rather than files, but doesn't explicitly contrast with 'query'.

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

Usage Guidelines2/5

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

The description provides no guidance on when to use this tool versus alternatives like 'upload_file' or 'query'. It mentions the tool's purpose but offers no context about prerequisites, appropriate scenarios, or exclusion criteria.

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

upload_fileB

Upload a file to the FileSearchStore for RAG indexing. The file will be processed and made searchable.

ParametersJSON Schema
NameRequiredDescriptionDefault
filePathYesAbsolute path to the file to upload (e.g., /path/to/document.pdf)
mimeTypeNoMIME type of the file (e.g., application/pdf, text/markdown). Auto-detected if not provided.
displayNameNoDisplay name for the file in the store. Uses filename if not provided.
metadataNoCustom metadata as key-value pairs. Values can be strings or numbers. Maximum 20 entries per document. Example: {"category": "guide", "year": 2025}

TDQS

B3.4/5.0
Behavior2/5

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

With no annotations provided, the description carries the full burden of behavioral disclosure. It mentions that the file 'will be processed and made searchable,' but lacks details on permissions, rate limits, error handling, or what 'processed' entails (e.g., indexing time, supported file types). This is inadequate for a mutation tool with zero annotation coverage.

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 front-loaded with the core purpose in the first sentence and adds a second sentence for behavioral context, with no wasted words. It is appropriately sized and efficiently structured.

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

Completeness2/5

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

For a mutation tool with no annotations and no output schema, the description is incomplete. It lacks details on behavioral traits (e.g., auth needs, processing behavior), error cases, and what the tool returns, leaving significant gaps for an AI agent to understand proper invocation.

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 fully documents all parameters. The description does not add any additional meaning or context beyond what the schema provides, such as explaining interactions between parameters or usage examples, resulting in a baseline score of 3.

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 specific action ('Upload a file') and the target resource ('to the FileSearchStore for RAG indexing'), distinguishing it from sibling tools like 'query' and 'upload_content' by specifying the file upload context for search indexing.

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 usage for making files searchable via RAG indexing, but it does not explicitly state when to use this tool versus alternatives like 'upload_content' or provide any exclusions or prerequisites for usage.

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. Dates show when Glama detected each change.

  1. 3 tool updates
    • First observedquery
    • First observedupload_content
    • First observedupload_file
Install Server
RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

TDQS

A3.5/5.0
Disambiguation5/5

Each tool has a clearly distinct purpose: querying for answers, uploading text content, and uploading files. There is no overlap in functionality, and an agent can easily tell them apart based on their specific roles in the RAG workflow.

Naming Consistency5/5

All tool names follow a consistent verb_noun pattern (query, upload_content, upload_file) with clear, descriptive terms. The naming is uniform and predictable, making it easy for agents to understand and use the tools.

Tool Count4/5

With 3 tools, the server is well-scoped for a RAG system, covering core operations: querying, uploading text, and uploading files. It is slightly lean but reasonable, as it handles the essential workflow without unnecessary complexity.

Completeness4/5

The tool set covers the main RAG operations: ingestion (uploading content/files) and retrieval (querying). Minor gaps might include tools for managing or deleting uploaded content, but the core functionality is complete for basic RAG use cases.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Connectors

Related MCP Servers

  • Chalee MCP RAG

    PrettyKing
    F
    license
    Not graded
    quality
    D
    maintenance
    Enables RAG (Retrieval-Augmented Generation) capabilities with document processing, vector storage, and intelligent Q\&A using OpenAI embeddings and semantic search.
    -

  • MCP RAG Server

    0xrdan
    A
    license
    Not graded
    quality
    C
    maintenance
    Enables AI assistants to search and retrieve information from your knowledge base using RAG (Retrieval-Augmented Generation) with hybrid search, document indexing, and ChromaDB vector storage.
    48
    MIT

  • knowledge_mgmt

    mmaun
    A
    license
    Not graded
    quality
    D
    maintenance
    Enables uploading, organizing, and semantically searching documents with support for various file types and embedding providers.
    30
    MIT

  • notebooklm-api

    mcpware
    A
    license
    A
    quality
    D
    maintenance
    Enables document Q&A with grounded citations using the official Gemini File Search API. Upload documents, create stores, and ask questions that return exact passage citations.
    7
    17
    2
    MIT

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/masseater/gemini-rag-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server