MCP Server
Connect Claude, Cursor, Codex, or any MCP client to the Roboflow MCP server, and the tools it exposes.
Work on your Roboflow projects together with AI. Connect Claude Code (or any MCP-compatible agent) to your workspace. It can create projects, upload data, train models, build Workflows, and guide you through the visual steps in the Roboflow UI. You handle what you're best at (seeing, labeling, judging results), your agent handles the rest.
Demo
Adding MCP
The Roboflow MCP server uses OAuth for authentication - no API key needed. You'll be prompted to sign in to Roboflow on first use.
Claude Connector (Recommended)
Add Roboflow as a connector in your Claude account. Once connected, it works everywhere - Claude.ai, Claude Desktop, and Claude Code.
Add Roboflow to Claude - click the link, confirm, and you're done. You'll be prompted to sign in to Roboflow with OAuth on first use.
Claude Code CLI
Cursor
Install the Roboflow plugin from the Cursor marketplace, or run /add-plugin roboflow in Cursor. Either installs the MCP server along with Roboflow's skills. Sign in to Roboflow with OAuth on first use.
To configure the server manually instead, add this to Cursor's MCP config (~/.cursor/mcp.json):
Codex
Add this to ~/.codex/config.toml:
Connecting MCP Gateways (Pre-Registered Credentials)
Most MCP clients (Cursor, Claude Desktop, VS Code, Claude Code) register automatically using Dynamic Client Registration. Some platforms require a pre-registered client_id and client_secret instead - for example, Azure AI Foundry agents, Microsoft Copilot Studio, and TrueFoundry AI Gateway.
To connect one of these platforms:
Go to Workspace Settings > Developer in the Roboflow dashboard
Click "Create OAuth App" and fill in a name, redirect URI (matching your gateway's callback URL), and allowed scopes
Set "Token endpoint authentication" to match your gateway:
client_secret_basic(HTTP Basic header, used by Azure and TrueFoundry) orclient_secret_post(secret in the form body)Copy the Client ID and Client Secret (shown once)
In your gateway's connector form, enter:
MCP Server URL
https://mcp.roboflow.com/mcp
Authorization URL
https://app.roboflow.com/oauth/authorize
Token URL
https://app.roboflow.com/oauth/token
Discovery (OAuth AS metadata)
https://app.roboflow.com/.well-known/oauth-authorization-server
Scopes
Space-separated list (ex: workspace:read project:read model:infer offline_access)
Include offline_access in your scope list if the gateway supports refresh tokens.
For the full scope catalog, see Available Scopes.
Tools
The Roboflow MCP server exposes the following tools, grouped by category. Tools are namespaced as mcp__<server>__<tool_name> by your MCP client (e.g. mcp__roboflow__projects_list).
Agent: chat with the Roboflow agent and build workflows
agent_chat
Chat with the Roboflow AI agent (Roboflow Q&A, advanced Workflow building, CV solution planning).
agent_conversation_get
Get a single agent conversation with its full message history.
agent_conversations_list
List Roboflow agent conversations in the workspace.
agent_workflow_publish
Publish the latest agent-edited draft of a workflow.
Projects: manage projects in your workspace
projects_list
List all projects in the workspace associated with the API key.
projects_get
Get detailed info about a project including versions, classes, splits, and trained models.
projects_create
Create a new computer vision project (detection, classification, segmentation, or keypoint).
projects_fork
Enqueue an async fork of a public Universe project into your workspace.
projects_health
Get the dataset health check for a project.
create_project_app
Open a Prefab UI so the human can create a Roboflow project (calls projects_create on confirm).
Images: upload and search project images
images_search
Search for images inside a project with filters for tag, class, batch, annotation job, and visual similarity.
images_prepare_upload
Get a pre-built upload URL to upload a single image to a project via curl.
images_prepare_upload_zip
Prepare a signed URL to bulk-upload a zip of images (and optional COCO/YOLO/VOC/folder annotations).
images_upload_zip_status
Check the status of a zip upload task started with images_prepare_upload_zip.
Annotation: save annotations and run autolabeling
annotations_save
Save an annotation (XML, JSON, or text) for an existing image.
autolabel_start
Start a hosted auto-label job over a batch of images.
autolabel_job_get
Get per-subjob status and progress for an auto-label job.
Batch: create and manage labeling jobs
annotation_jobs_create
Create an annotation job to assign a batch of images to a labeler and reviewer.
annotation_batches_list
List upload batches in a project.
annotation_batches_get
Get details about a specific batch including image count and status.
Versions: create and export dataset versions
versions_get
Get info about a dataset version including splits, preprocessing, augmentation, and model metrics.
versions_generate
Create a new dataset version with optional preprocessing and augmentation settings.
versions_export
Check or trigger a dataset export for a version in a specific format (e.g. coco, yolov8, jsonl).
Models: train models and monitor training progress
models_list
List trained models associated with a project.
models_get
Get details for a trained model.
models_train
Start training a model (RF-DETR, YOLO, etc.) on a dataset version.
models_infer
Run hosted inference on an image using a trained model and return JSON predictions.
models_get_training_status
Get the training progress and metrics (epoch, mAP, loss) for a dataset version.
models_star_nas
Star or unstar a NAS-trained model.
trainings_get_results
Get the run-level training results bundle.
trainings_stop
Request an early stop on an in-flight training run.
trainings_cancel
Cancel an in-flight training run.
8. Model Evaluations: view mAP, confusion matrices, and per-class performance
model_evals_list
List model evaluations in the current workspace.
model_evals_get
Get the top-level summary for a single model evaluation.
model_evals_get_map_results
Get per-split mAP results for an eval.
model_evals_get_confusion_matrix
Get the confusion matrix for an eval.
model_evals_get_confidence_sweep
Get the precision/recall/F1 confidence sweep for an eval.
model_evals_get_performance_by_class
Get per-class performance metrics for a single split.
model_evals_get_image_predictions
Get per-image prediction stats for an eval (paginated).
model_evals_get_vector_analysis
Get UMAP + HDBSCAN clustering of image embeddings for an eval.
model_evals_get_recommendations
Get the LLM-generated recommendations for an eval, if available.
9. Workflows: build and run inference pipelines
workflows_list
List saved workflows in the current workspace.
workflows_get
Get details for a saved workflow.
workflows_create
Create and save a new Workflow in the workspace from a JSON definition.
workflows_update
Update an existing saved Workflow's name and definition.
workflows_run
Execute a saved Workflow by ID on one or more images.
workflow_blocks_list
List all available Workflow blocks with a short summary of each.
workflow_blocks_get_schema
Get the full schema (properties, required fields) of a specific Workflow block.
workflow_specs_validate
Validate a Workflow JSON definition without executing it.
workflow_specs_run
Execute a Workflow directly from an inline JSON definition without saving it first.
Devices & Streams: manage edge devices and video streams
devices_list
List devices registered in the workspace.
devices_get
Get a single device by id.
devices_create
Provision a new v2 device.
devices_get_config
Get the device's current runtime configuration.
devices_get_default_config
Get the workspace's default device configuration.
devices_get_config_history
List prior configuration revisions, newest first.
devices_update_config
Update the device's runtime configuration.
devices_get_logs
Fetch device logs from Elasticsearch.
devices_get_events
List device and stream lifecycle events.
devices_get_telemetry
Get aggregated hardware metrics (cpu, memory, disk, gpu).
devices_streams_list
List streams configured on the device.
devices_streams_get
Get a single stream by id.
Universe: search public datasets on Roboflow Universe
universe_search
Search Roboflow Universe for public datasets or models, with operators for class, image count, project type, and sort.
universe_search_app
Open a Prefab Universe UI for search, visual comparison, image skim, and fork - the human picks the dataset.
universe_dataset_images_search
Search images inside a public Universe dataset URL.
Misc: poll long-running tasks and send feedback
async_tasks_get
Poll an async task by id (e.g. project forks). Poll every 5 seconds; processing may take up to 30 seconds to start.
meta_feedback_send
Report a bug, missing feature, UX friction, or documentation issue back to the Roboflow team.
Last updated
Was this helpful?