Roboflow MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ROBOFLOW_API_KEY | Yes | Your Roboflow API key (required) | |
| ROBOFLOW_API_URL | No | Override the base API URL | https://api.roboflow.com |
| ROBOFLOW_WORKSPACE | No | Default workspace ID | API key owner's workspace |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| roboflow_api_statusA | Check Roboflow API connectivity and key validity. Use as a first call to verify setup. |
| roboflow_universe_searchA | Search Roboflow Universe for existing models and datasets. Use before building a custom model to check if a solution already exists. |
| roboflow_project_listA | List projects in the authenticated Roboflow workspace. |
| roboflow_pretrained_listA | List Roboflow's curated pre-trained APIs for common vision tasks (OCR, people detection, PPE, license plates, etc.). These are production-ready, zero-setup endpoints. |
| roboflow_rapid_createA | Create a model using Roboflow Rapid (prompt-based, no training data required). Describe what to detect in natural language. |
| roboflow_inference_runA | Run object detection or segmentation inference on a Roboflow model. Accepts image as URL, base64, or local file path. |
| roboflow_inference_classifyA | Run classification inference on a Roboflow model. Returns top-K class predictions with confidence scores. |
| roboflow_workflow_listA | List available Roboflow Workflows in a workspace. |
| roboflow_workflow_runA | Execute a Roboflow Workflow. Workflows chain multiple models and logic steps into a single pipeline. |
| roboflow_upload_imageA | Upload an image to a Roboflow project dataset for training. Specify the dataset split (train/valid/test). |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 10 tools
Each tool targets a distinct operation: status checking, searching, listing different resource types, creating via Rapid, running different inference types, executing workflows, and uploading images. The only potential overlap between inference_run and inference_classify is clearly differentiated by model task type.
All tools share the 'roboflow_' prefix, but the suffix pattern is inconsistent: some use noun_verb (universe_search, inference_run), some use noun_noun (project_list, api_status), and one uses verb_noun (upload_image). This mix of patterns makes it less predictable than a uniform convention.
With 10 tools, the server is well-scoped, covering the core Roboflow workflows without being bloated. Each tool serves a clear purpose and the count fits the intended functionality.
The set covers discovery, inference, workflow execution, and data upload, but lacks project management operations like create/update/delete project or dataset. This creates notable gaps for users who need to set up new projects outside of Rapid creation.