Skip to main content
Glama
srigi

Google Images Search MCP

by srigi
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Hybrid

🖼️ Google Images Search MCP

Search. See. Save. – The only MCP that shows Google Images results directly in your chat as actual images, not just links!

Unlike other tools that return text descriptions or links, this MCP displays search results as actual images in your conversation. Works with Cline and Windsurf.

Perfect for getting visual inspiration, finding assets for projects, or quickly browsing images without leaving your IDE.

Demo

Demo

Related MCP server: DuckDuckGo MCP Server

Usage & Requirements

Requirements

  • Node.js v20+

  • Google Cloud Platform account (free tier sufficient)

  • Compatible IDE: Cline or Windsurf

Setup Guide

Follow these steps to set up this MCP:

Step 1: create Google Custom search engine

  1. Go to Google Programmable Search Engine

  2. Click "Add" to create a new search engine

  3. In "Sites to search", enter * (asterisk) to search the entire web

  4. Give your search engine a name (e.g., "My Image Search")

  5. In the search engine settings, enable "Image search" option

  6. Click "Create"

  7. Copy the Search engine ID - you'll need this for configuration

Step 2: Get Google API Key

  1. Go to Google Cloud Console API Credentials

  2. If you don't have a project, create one

  3. Click "+ CREATE CREDENTIALS" → "API key"

  4. Copy the generated API key

  5. (Optional) Restrict the API key to "Custom Search API" for security

Step 3: Configure MCP in Your IDE

Add the following configuration to your MCP settings:

{
  "mcpServers": {
    "googleImagesSearch": {
      "command": "npx",
      "args": ["-y", "@srigi/mcp-google-images-search"],
      "env": {
        "API_KEY": "your-google-api-key-here",
        "SEARCH_ENGINE_ID": "your-search-engine-id-here"
      },
      "autoApprove": ["search_image", "persist_image"]
    }
  }
}

Replace your-google-api-key-here and your-search-engine-id-here with the values from steps 1 and 2.

Usage Example

Here's how to use the MCP once configured:

  1. Search for images: Ask your AI assistant to search for images

    Find 5 images of F-22

  2. Get more results: Request additional search results

    Find 5 more images

  3. Save an image: Ask to save a specific result to your project

    Save the 3rd image to the "assets" folder

The MCP will display the search results as actual images in your chat history, and you can easily save any of them to your local project directory.

🚀 Development

Want to contribute? Great! Quality contributions are welcomed.

Requirements: Node.js v20+, direnv, PNPM v10

Quick start:

  1. Copy .envrc (example) to .envrc and add your Google API credentials

  2. Run pnpm install && pnpm dev

  3. Update your MCP configuration for development:

{
  "mcpServers": {
    "googleImagesSearch": {
      "command": "node",
      "args": ["/absolute/path/to/project/src/index.js", "--debug", "--pretty-print"],
      "env": {
        "API_KEY": "your-google-api-key-here",
        "SEARCH_ENGINE_ID": "your-search-engine-id-here"
      },
      "autoApprove": ["search_image", "persist_image"]
    }
  }
}

Debug and logging options:

  • No --debug flag provided - logging is disabled

  • --debug without argument - enable logging into debug.log in current working directory the running MCP

  • --debug /absolute/path/to/debug.log - enable logging and write logs to the specified absolute path

  • --pretty-print - Format logs for better readability (formatted JSON output)

note: only absolute path is allowed when providing a path to --debug. Relative paths will cause the server to exit with an error!

Writing TypeScript code:

The pnpm dev auto-recompile all changes to .ts files. But you must reload your MCP server manually for changes to be reflected in your IDE.

Use pnpm dev:inspector for interactive testing in the browser.

🛠️ Available Tools

🔍 search_image - Find images using Google's vast database

  • query (required) - What you're looking for

  • count (1-10, default: 2) - How many results

  • safe ('off'/'medium'/'high') - Filter level

  • startIndex - For pagination

💾 persist_image - Download and save images to your project

  • url (required) - Image URL to download

  • targetPath (required) - Where to save it (folder or full path)

Security features: Path validation, MIME type checking, 10MB size limit, supports all major image formats (JPEG, PNG, GIF, WebP, SVG, etc.)

🔗 Links: GitHub • NPM • Issues

Available Tools

2 tools
persist_imageA

Store image at URL to folder relative to current workspace. If targetPath does not exist, the tool will create it automatically.

ParametersJSON Schema
NameRequiredDescriptionDefault
urlYesURL of the image
targetPathYesFolder where to save the image (relative to the current workspace)
workspacePathYesThe current workspace absolute path

TDQS

A4.1/5.0
Behavior3/5

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

Without annotations, the description carries the full burden. It discloses a key behavior (auto-creation of targetPath), but does not mention what happens if the file already exists (overwrite?) or what occurs if the URL is invalid. This leaves important side effects unspecified.

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?

Two sentences, zero waste. The key information is front-loaded, and the conditional behavior follows naturally. No redundancy or filler.

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?

The description covers the core operation and auto-creation, but omits details about how the image file is named (e.g., derived from URL) and potential error handling. Given the tool's simplicity and full schema coverage, these gaps could mislead an agent about the exact output.

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

Parameters4/5

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

Schema coverage is 100%, so baseline is 3. The description adds value beyond the schema by explicitly stating that targetPath need not pre-exist and will be created automatically. This clarifies the parameter's semantics beyond the schema's 'Folder where to save the image'.

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 action ('Store image at URL') and the destination ('folder relative to current workspace'), making the tool's purpose unambiguous. It distinguishes from sibling 'search_image' by focusing on persistence rather than discovery.

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?

The description gives clear context: it saves an image to a workspace-relative folder and automatically creates the target directory if missing. It does not explicitly mention alternatives or exclusions, but the 'relative to current workspace' and 'auto-create' instructions provide sufficient usage context for most cases.

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

search_imageC

Search the image(s) online

ParametersJSON Schema
NameRequiredDescriptionDefault
countNoNumber of results to return (1-10, default: 2)
queryYesSearch query for images
safeNoSafe search setting (default: off)
startIndexNoStarting index of next search result page (not needed for initial search request)

TDQS

C2.5/5.0
Behavior2/5

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

With no annotations, the description carries the full burden of behavioral disclosure. It does not mention safe search settings, pagination, result limits, or any side effects. The agent is left unaware of how the search behaves or what the response contains.

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 a single short sentence with no wasted words. It is front-loaded with the core action, but it is too sparse to be truly helpful, making it concise but under-specified.

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

Completeness1/5

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

Given the tool has 4 parameters and no output schema, the description is severely incomplete. It fails to mention the result format, pagination behavior, or safe search implications, leaving critical gaps for an agent to invoke and interpret the tool correctly.

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 input schema fully documents all parameters. The description adds no semantic detail beyond the schema, but the baseline of 3 applies because the schema carries the burden effectively.

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

Purpose3/5

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

The description 'Search the image(s) online' identifies a clear verb and resource, but it is vague about the scope and output. It does not differentiate from the sibling tool 'persist_image' beyond the search action, and 'online' adds little specificity.

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?

There is no guidance on when to use this tool versus alternatives. The sibling 'persist_image' exists, but the description does not mention any selection criteria or exclusions, leaving the agent without context for choosing this tool.

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. 2 tool updatesv0.3.0
    • Changedpersist_image2 fields changed
      • addedInput schema / $schema
        Added value: +"http://json-schema.org/draft-07/schema#"
      • addedInput schema / additionalProperties
        Added value: +false
    • Changedsearch_image2 fields changed
      • addedInput schema / $schema
        Added value: +"http://json-schema.org/draft-07/schema#"
      • addedInput schema / additionalProperties
        Added value: +false
  2. 2 tool updatesv1.0.0
    • First observedpersist_image
    • First observedsearch_image
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

B3.3/5.0

Scored across 2 tools

Disambiguation5/5

The two tools have clearly distinct purposes: search_image retrieves images online, persist_image saves a specific image from a URL. There is no overlap or ambiguity between them.

Naming Consistency5/5

Both tool names follow a consistent verb_noun pattern (search_image, persist_image) using snake_case, which is predictable and clear.

Tool Count3/5

With only 2 tools, the set is minimal for a search-and-persist workflow. While it covers the core actions, it feels slightly thin compared to typical server scopes of 3-15 tools.

Completeness4/5

The tools cover the essential workflow: search for images and persist a chosen image. No obvious gaps for the stated purpose, though additional functions like listing persisted images or metadata retrieval could be useful.

Maintenance

ActivityInactive
ResponsivenessUnresponsive

Resources

Related MCP Connectors

Related MCP Servers

Appeared in Searches