Mapillary MCP for AI. Analyze street view imagery, object detections, and map features.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Connect to your AI in seconds.
Mapillary connects your AI client to the world's largest street-level imagery platform. You search for images, map features (like traffic signs), and object detections across any area globally.
It gives you image metadata—GPS coordinates, capture dates, and compass angles—so you can pinpoint exactly what was captured where.
What your AI can do
Get detection value
Gets specific details—type, GPS coordinates, and association—for a single detected object.
Get image
Returns comprehensive metadata for one image ID, including capture time, location, and sequence ID.
Get image detections
Lists all detected objects within a specific image, providing their types, geometry, and confidence scores.
You ask your agent to find street-level photos or sequences along a specific route using coordinates.
Your agent searches for map features—like stop signs, speed limits, and road markings—in a defined area, useful for planning or auditing.
You instruct your agent to list every detected object (e.g., streetlights, vehicles) found within a specific image ID.
The server returns the full details for an image, including its capture date, precise GPS coordinates, and angle.
You retrieve granular data on a single object detection, showing its specific type, value, and exact location within the photo.
Ask an AI about this
Waiting for input…
Mapillary MCP Server: 7 Tools for Geospatial Analysis
This server provides granular access to street-level imagery data. Use the tools to search areas, retrieve image metadata, and detect specific objects or map features.
Make your AI actually useful.
Add this MCP to Claude, Cursor, or Windsurf and your AI stops guessing. It gets real tools to look things up, take action, and handle the stuff you keep doing by hand.
Start using Mapillary on VinkiusGet Detection Value
Gets specific details—type, GPS coordinates, and association—for a single detected object.
Get Image
Returns comprehensive metadata for one image ID, including capture time, location...
Get Image Detections
Lists all detected objects within a specific image, providing their types, geometry...
Get Map Features
Searches for static map elements like traffic signs or road markings across an area...
Get Sequence
Retrieves metadata and linked images for a continuous set of photos captured along a...
Search Images
Finds image IDs, coordinates, and capture dates by searching within a defined geographic bounding box.
Search Sequences
Locates entire photo sequences (routes) that pass through a specified geographical area.
Security and governance baked right in.
Pick your AI client below to get set up. Just create a Vinkius account, subscribe, and you're instantly up and running. We handle the entire backend infrastructure, delivering out-of-the-box support for HTTPS Streamable, SSE, and OAuth2—zero messy routing required.
Claude AI
Open Claude Settings
Go to claude.ai, click your profile icon, then navigate to Customize → Connectors.
Add Custom Connector
Click the "+" button and select Add custom connector. Paste your Vinkius endpoint URL:
https://edge.vinkius.com/[YOUR_TOKEN_HERE]/mcp
Replace [YOUR_TOKEN_HERE] with your token
from cloud.vinkius.com. For OAuth-protected servers, expand
Advanced settings to add credentials.
Start a conversation
Open a new chat. The Mapillary integration is available immediately — no restart needed.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Mapillary, then connect any of our 5,000+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 5,000+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Mapillary. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This connection provides 7 powerful capabilities that interface natively with Claude, ChatGPT, Cursor, and other compatible AI platforms. No middleware. No custom integration required.
Manually checking street view data takes hours of clicking and cross-referencing tabs.
Right now, if you're auditing traffic signs or object placement, you pull up a map service. You draw a box, see hundreds of potential points, and then you have to open each one individually. You copy the coordinates into a spreadsheet, open another window to check the date, and manually verify if the sign was even visible in that specific photo.
With this MCP server, you tell your agent: 'Find all speed limit signs between these two points.' The agent runs `get_map_features` across the area. You get a structured list of results—GPS coordinates, feature type, and confidence score—all in one go.
Mapillary MCP Server: Getting object detections from visual data.
Manual processes often stop at 'I see an object.' You get a vague detection box. You don't know if it’s a streetlight, or just a pole—and you certainly don't have the exact GPS location of the base, only the bounding box.
The Mapillary MCP Server gives you that specificity. You run `get_image_detections` and immediately get the value, type, geometry, *and* confidence score for every object found in a single image. It turns vague observation into auditable data points.
What your AI can actually do with this
Mapillary lets your agent access the world’s street-level imagery data through natural conversation.
This server exposes several tools for geospatial analysis:
- Search Images: Pinpoint images using a geographic bounding box.
- Search Sequences: Find connected image sets (routes) that pass through an area.
- Map Features: Search for specific static elements like traffic signs or road markings by location.
- Image Detection: Get object detections and detailed values from specific images, linking the detection to its geometry and confidence score.
019d8455-08c0-708f-b8cc-d4f5194b9148 
Here's how it actually works
The bottom line is you get structured access to massive amounts of geospatial data that would otherwise require dedicated GIS software.
First, your agent uses search_images or get_map_features with coordinates to narrow down the area of interest.
Next, you pass a specific image ID or sequence ID to get_image or get_image_detections to retrieve the raw data and associated objects.
Finally, your agent processes the output using get_detection_value or get_sequence to extract actionable metadata—like a speed limit value or GPS path.
Who is this actually for?
Urban planners, environmental researchers, and civil engineers use this. They need to audit physical infrastructure—traffic signs, road markings, etc.—across large areas without manual site visits. If you're tired of relying on vague satellite views for ground truth data, this is for you.
Uses get_map_features and search_images to inventory traffic signs or analyze road markings across a neighborhood grid.
Combines get_image_detections with GPS data from get_image to build environmental models based on captured objects over time.
Uses the full suite of tools, especially object detection and sequence search, to validate perception stack performance against real-world visual evidence.
What Changes When You Connect
Find objects across wide areas: Instead of checking one photo at a time, use get_map_features to search for all traffic signs or road markings in an entire quadrant.
Understand the journey's context: Use search_sequences and then get_sequence to track how conditions change along a specific route path over time.
Deep dive into single images: If you find an image ID using search_images, run get_image_detections immediately. This lists every object the system found, giving confidence scores for each one.
Pinpoint exact object data: Don't just know a car was there; use get_detection_value to get that specific detection's type, GPS coordinates, and image link.
Get full photo context: Use get_image on any ID to pull back the original capture date, compass angle, and organization data—the whole metadata package.
See it in action
Auditing Speed Limits
A planner needs to check if all speed limits are posted on a major street. They use get_map_features over the street's bounding box. The agent filters results for 'speed limit sign' and gets GPS coordinates and confidence scores, confirming compliance quickly.
Tracking Changes in Urban Areas
A researcher wants to know what objects were present in a park six months ago versus today. They use search_sequences to find the path, then compare the object lists from get_image_detections for corresponding image IDs.
Validating Autonomous Driving Paths
A developer simulates a driving route using coordinates and calls search_images. For key intersections, they run get_image_detections to confirm the agent's perception stack can identify everything from stop signs to pedestrian crossings.
Forensic Image Analysis
Given a specific image ID, the user needs to know exactly what was captured. They call get_image, which provides the full metadata (GPS, date) and then call get_image_detections for the list of objects.
The honest tradeoffs
Searching blindly by keyword
Asking the agent: 'Show me pictures of traffic signs in London.' This is too vague and misses critical spatial data.
You must define a physical boundary. First, use search_images with the correct bounding box (e.g., '-0.15,51.50,-0.10,51.52'). Then, run get_map_features to specifically filter for 'traffic sign' within that area.
Treating all photos the same
Calling get_image_detections without first knowing if an image ID is valid or what its capture date was. This leads to missing context.
Always start by calling get_image(image_id) to confirm the metadata, location, and sequence details before attempting any deep detection calls.
Missing the route context
Only searching for images using search_images when the user actually wants a continuous path view. This gives isolated points.
If you need multiple connected photos, use search_sequences first to find the sequence ID. Then pass that ID to get_sequence for full context.
When It Fits, When It Doesn't
Use this server if your job requires physical ground truth data: analyzing infrastructure, detecting objects in place, or tracing routes using coordinates. You must use it when you need more than just a general map view—you need the actual image-level detections and metadata.
Don't use this if you only need to know which street is near another (use standard mapping APIs for that). Also, don't rely on it if your goal is merely generating conceptual data; you must always have specific coordinates or bounding boxes to start. If you only want a list of general points of interest without detection capabilities, other map services might suffice.
To get the most out of this, you need an orchestration layer that chains calls: search_sequences -> get_sequence -> (loop through image IDs) get_image_detections. This complex workflow is where the power lies.
Questions you might have
How do I search street view images using `search_images`? +
You pass the function your bounding box parameters: min_lon, min_lat, max_lon, max_lat. The server returns a list of image IDs, coordinates, and capture dates that fall within that specific rectangle.
What is the difference between `search_images` and `search_sequences`? +
search_images finds individual, isolated photos. search_sequences finds continuous paths or routes composed of multiple connected images; you'll get a sequence ID instead.
Can I find traffic signs using the `get_map_features` tool? +
Yes, that’s exactly what it does. You query get_map_features by area and filter for 'traffic sign' or other road markings to build an inventory.
What fields can I request when using `get_image`? +
You pass a fields parameter. You can request specific metadata like geometry, compass angle, captured_at, sequence ID, or various thumbnail URLs to get exactly the data you need.
How do I authenticate my agent to use the `get_image` tool? +
You first subscribe to the Mapillary server and provide your unique access token. Once connected, you can call get_image with specific coordinates or IDs to pull detailed metadata about any street-level photo.
What is the best way to handle high volumes of data when running a query using `search_images`? +
Be mindful of your rate limits and bounding box size. For large areas, break your search into smaller, manageable geographical sections. This prevents quota overruns and keeps results stable.
If I get a list of detections for an image using `get_image_detections`, how do I get more detail on one specific object? +
You pass the unique detection ID to the get_detection_value tool. This gives you granular information, including the object's exact type, confidence score, and precise GPS coordinates.
When using `get_map_features`, what format must I use for my search area? +
You need to provide your search area as a geographical bounding box. This is formatted as four comma-separated values: min_lon, min_lat, max_lon, and max_lat.
How do I get a Mapillary access token? +
Sign up at mapillary.com/developer and create an application to get your access token.
What is a bounding box? +
A bounding box defines a geographic area using min_longitude, min_latitude, max_longitude, max_latitude. Example: "-0.15,51.50,-0.10,51.52" covers central London.
What map features are available? +
Mapillary detects traffic signs (stop, speed limit, yield, etc.), road objects (fire hydrants, mailboxes, etc.) and road markings. Use get_map_features to search by area.
We've already built the connector for Mapillary. Just plug in your AI agents and start using Vinkius.
No hosting. No infrastructure. No complex setup.
All 7 tools are live and waiting.
You're up and running in seconds.
Gemini Vinkius gives your AI agents access to the full catalog of app connectors, all fully managed, secure, and enterprise-ready. One subscription, every tool you need.
Built, hosted, and secured by Vinkius. You just connect and go.