Prefect MCP Server
The Prefect MCP Server enables AI assistants to interact with Prefect's workflow orchestration platform through natural language via the Model Context Protocol (MCP), providing comprehensive management and monitoring capabilities through 45+ API endpoints.
Core Capabilities:
Flow Management: List, filter, view, and delete flows; query flow runs with extensive filtering (state, deployment, tags, time ranges); control flow run lifecycle (create, restart, cancel, delete, set states); retrieve flow run details and logs
Deployment Operations: List, filter, view, update, and delete deployments; create flow runs from deployments with custom parameters, names, tags, and idempotency keys; manage deployment schedules (view, set cron/interval-based schedules, pause, resume); configure work queue assignments
Task Run Monitoring: List and filter task runs by name, state, tags, and time ranges; retrieve task run details and logs; query task runs by flow run; manually set task run states with optional messages
Work Queue Operations: Create, list, update, pause, resume, and delete work queues; manage concurrency limits and configuration; filter by name or paused status
Variable Management: Create, read, update, and delete variables with support for string, dict, list, or other JSON-serializable values; manage variable tags; filter by name with pagination
Block Management: List available block types with slug-based filtering; retrieve block type schemas; query, retrieve, and delete block documents (configuration instances)
Workspace Management: List accessible workspaces; get current workspace context; retrieve workspace details by ID or handle
Artifact Management: Perform CRUD operations on Prefect artifacts
Automation Control: Create, view, update, delete, pause, and resume automations
System Monitoring: Check Prefect server health status and connectivity; create and retrieve logs for debugging
Key Features: Comprehensive filtering on most operations (tags, names, states, dates), pagination support with limit/offset, UUID-based precise resource targeting, tag-based organization, and state management with custom messages.
Allows AI assistants to interact with Prefect through natural language, providing access to flow management, flow run management, deployment management, task run monitoring, work queue management, block management, variable management, and workspace information.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Prefect MCP Server show me my recent flow runs and their statuses"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Deprecated 27-Nov-2025
I've personally moved my efforts to a more generic OpenAPI spec based MCP: https://github.com/allen-munsch/yas-mcp
Additionally, there is actually an official beta release by prefect over here: https://pypi.org/project/prefect-mcp/
Prefect MCP Server
A Model Context Protocol (MCP) server implementation for Prefect, enabling AI assistants to interact with Prefect through natural language.
Note: The official Prefect MCP server is available here. This is a community implementation.
๐ Quick Start
docker compose upRelated MCP server: n8n MCP Server
๐ฆ Installation
pip Installation
pip install mcp-prefectFrom Source
git clone https://github.com/allen-munsch/mcp-prefect
cd mcp-prefect
pip install -e .Manual Run
PREFECT_API_URL=http://localhost:4200/api \
PREFECT_API_KEY=your_api_key_here \
MCP_PORT=8000 \
python -m mcp_prefect.main --transport http๐ ๏ธ Features
โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ โ
โ _ __ ___ _____ __ __ _____________ ____ ____ โ
โ _ __ ___ .'____/___ ______/ /_/ |/ / ____/ __ \ |___ \ / __ \ โ
โ _ __ ___ / /_ / __ `/ ___/ __/ /|_/ / / / /_/ / ___/ / / / / / โ
โ _ __ ___ / __/ / /_/ (__ ) /_/ / / / /___/ ____/ / __/_/ /_/ / โ
โ _ __ ___ /_/ \____/____/\__/_/ /_/\____/_/ /_____(*)____/ โ
โ โ
โ โ
โ FastMCP 2.0 โ
โ โ
โ โ
โ ๐ฅ๏ธ Server name: MCP Prefect 3.6.1 โ
โ ๐ฆ Transport: STDIO โ
โ โ
โ ๐๏ธ FastMCP version: 2.12.3 โ
โ ๐ค MCP SDK version: 1.14.1 โ
โ โ
โ ๐ Docs: https://gofastmcp.com โ
โ ๐ Deploy: https://fastmcp.cloud โ
โ โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
[11/11/25 02:08:06] INFO Starting MCP server 'MCP Prefect 3.6.1' with transport 'stdio' server.py:1495
โ
Initialized successfully
Server: MCP Prefect 3.6.1 1.14.1
๐ Listing tools...
๐ฏ FOUND 64 TOOLS:
================================================================================
๐ ARTIFACT (6 tools)
๐ง create_artifact
๐ง delete_artifact
๐ง get_artifact
๐ง get_artifacts
๐ง get_latest_artifacts
๐ง update_artifact
๐ AUTOMATION (7 tools)
๐ง create_automation
๐ง delete_automation
๐ง get_automation
๐ง get_automations
๐ง pause_automation
๐ง resume_automation
๐ง update_automation
๐ BLOCK (5 tools)
๐ง delete_block_document
๐ง get_block_document
๐ง get_block_documents
๐ง get_block_type
๐ง get_block_types
๐ DEPLOYMENT (8 tools)
๐ง delete_deployment
๐ง get_deployment
๐ง get_deployment_schedule
๐ง get_deployments
๐ง pause_deployment_schedule
๐ง resume_deployment_schedule
๐ง set_deployment_schedule
๐ง update_deployment
๐ FLOW (13 tools)
๐ง cancel_flow_run
๐ง create_flow_run_from_deployment
๐ง delete_flow
๐ง delete_flow_run
๐ง get_flow
๐ง get_flow_run
๐ง get_flow_run_logs
๐ง get_flow_runs
๐ง get_flow_runs_by_flow
๐ง get_flows
๐ง get_task_runs_by_flow_run
๐ง restart_flow_run
๐ง set_flow_run_state
๐ LOG (2 tools)
๐ง create_log
๐ง get_logs
๐ OTHER (1 tools)
๐ง get_health
๐ TASK (4 tools)
๐ง get_task_run
๐ง get_task_run_logs
๐ง get_task_runs
๐ง set_task_run_state
๐ VARIABLE (5 tools)
๐ง create_variable
๐ง delete_variable
๐ง get_variable
๐ง get_variables
๐ง update_variable
๐ WORK (13 tools)
๐ง create_work_queue
๐ง delete_work_queue
๐ง get_current_workspace
๐ง get_work_queue
๐ง get_work_queue_by_name
๐ง get_work_queue_runs
๐ง get_work_queues
๐ง get_workspace
๐ง get_workspace_by_handle
๐ง get_workspaces
๐ง pause_work_queue
๐ง resume_work_queue
๐ง update_work_queue
๐ TOTAL: 64 tools across 10 categories๐ฌ Example Interactions
AI assistants can help you with:
Flow Management
"Show me all my flows and their last run status"
"Create a new flow run for the 'data-processing' deployment"
"What's the current status of flow run 'abc-123'?"
Deployment Control
"Pause the schedule for the 'daily-reporting' deployment"
"Update the 'etl-pipeline' deployment with new parameters"
Infrastructure Management
"List all work pools and their current status"
"Create a new work queue for high-priority jobs"
Variable & Configuration
"Create a variable called 'api_timeout' with value 300"
"Show me all variables containing 'config' in their name"
Monitoring & Debugging
"Get the logs for the last failed flow run"
"Show me all running task runs right now"
๐ค Platform Integration
Claude Desktop
Add to claude_desktop_config.json:
{
"mcpServers": {
"prefect": {
"command": "mcp-prefect",
"args": ["--transport", "stdio"]
}
}
}Cursor MCP
{
"mcpServers": {
"prefect": {
"command": "mcp-prefect",
"args": ["--transport", "stdio"]
}
}
}Gemini CLI
gemini config set mcp-servers.prefect "mcp-prefect --transport stdio"Windsurf / Claude Code
{
"mcpServers": {
"prefect": {
"command": "mcp-prefect",
"args": ["--transport", "stdio"],
"env": {
"PREFECT_API_URL": "http://localhost:4200/api",
"PREFECT_API_KEY": "your_api_key_here"
}
}
}
}Generic MCP Client
{
"mcpServers": {
"prefect": {
"command": "mcp-prefect",
"args": ["--transport", "stdio"],
"env": {
"PREFECT_API_URL": "http://localhost:4200/api",
"PREFECT_API_KEY": "your_api_key_here"
}
}
}
}๐งช Development
Running Tests
pytest tests/ -vBuilding from Source
git clone https://github.com/allen-munsch/mcp-prefect
cd mcp-prefect
pip install -e .
python -m mcp_prefectAvailable Tools
46 toolscancel_flow_runC
Cancel a flow run.
Args: flow_run_id: The flow run UUID
Returns: Confirmation message
Inputs: flow_run_id (required).
create_flow_run_from_deploymentC
Create a flow run from a deployment.
Args: deployment_id: The deployment UUID parameters: Optional parameters to pass to the flow run name: Optional name for the flow run tags: Optional tags for the flow run idempotency_key: Optional idempotency key
Returns: Details of the created flow run
Inputs: deployment_id (required), idempotency_key, name, parameters, +1 more.
create_variableB
Create a variable.
Args: name: The variable name value: The variable value (can be string, dict, list, etc.) tags: Optional tags
Returns: Details of the created variable
Inputs: name (required), value (required), tags.
create_work_queueB
Create a work queue.
Args: name: The name for the work queue description: Optional description is_paused: Whether the queue should be paused upon creation concurrency_limit: Optional concurrency limit
Returns: Details of the created work queue
Inputs: name (required), concurrency_limit, description, is_paused.
delete_block_documentB
Delete a block document by ID.
Args: block_document_id: The block document UUID
Returns: Confirmation message
Inputs: block_document_id (required).
delete_deploymentC
Delete a deployment by ID.
Args: deployment_id: The deployment UUID
Returns: Confirmation message
Inputs: deployment_id (required).
delete_flowB
Delete a flow by ID.
Args: flow_id: The flow UUID
Returns: Confirmation message
Inputs: flow_id (required).
delete_flow_runC
Delete a flow run.
Args: flow_run_id: The flow run UUID
Returns: Confirmation message
Inputs: flow_run_id (required).
delete_variableC
Delete a variable by name.
Args: name: The variable name
Returns: Confirmation message
Inputs: name (required).
delete_work_queueC
Delete a work queue by ID.
Args: work_queue_id: The work queue UUID
Returns: Confirmation message
Inputs: work_queue_id (required).
get_block_documentB
Get a block document by ID.
Args: block_document_id: The block document UUID
Returns: Block document details
Inputs: block_document_id (required).
get_block_documentsB
Get block documents by block type.
Args: block_type_slug: The block type slug limit: Maximum number of block documents to return offset: Number of block documents to skip name: Filter by name pattern
Returns: A list of block documents with their details
Inputs: block_type_slug (required), limit, name, offset.
get_block_typeC
Get a block type by slug.
Args: slug: The block type slug
Returns: Block type details
Inputs: slug (required).
get_block_typesB
Get a list of block types with optional filtering.
Args: limit: Maximum number of block types to return offset: Number of block types to skip slug: Filter by slug pattern
Returns: A list of block types with their details
Inputs: limit, offset, slug.
get_current_workspaceC
Get the current workspace.
Returns: Details of the current workspace
Inputs: no named parameters.
get_deploymentC
Get details of a specific deployment by ID.
Args: deployment_id: The deployment UUID
Returns: Deployment details
Inputs: deployment_id (required).
get_deploymentsB
Get a list of deployments with optional filtering.
Args: limit: Maximum number of deployments to return offset: Number of deployments to skip flow_name: Filter by flow name name: Filter by deployment name tags: Filter by tags is_schedule_active: Filter by schedule active status work_queue_name: Filter by work queue name
Returns: A list of deployments with their details
Inputs: flow_name, is_schedule_active, limit, name, +3 more.
get_deployment_scheduleC
Get a deployment's schedule.
Args: deployment_id: The deployment UUID
Returns: Schedule details
Inputs: deployment_id (required).
get_flowC
Get details of a specific flow by ID.
Args: flow_id: The flow UUID
Returns: Flow details
Inputs: flow_id (required).
get_flow_runB
Get details of a specific flow run by ID.
Args: flow_run_id: The flow run UUID
Returns: Flow run details
Inputs: flow_run_id (required).
get_flow_runsB
Get a list of flow runs with optional filtering.
Args: limit: Maximum number of flow runs to return offset: Number of flow runs to skip flow_name: Filter by flow name state_type: Filter by state type (e.g., "RUNNING", "COMPLETED", "FAILED") state_name: Filter by state name deployment_id: Filter by deployment ID tags: Filter by tags start_time_before: ISO formatted datetime string start_time_after: ISO formatted datetime string
Returns: A list of flow runs with their details
Inputs: deployment_id, flow_name, limit, offset, +5 more.
get_flow_runs_by_flowB
Get flow runs for a specific flow.
Args: flow_id: The flow UUID limit: Maximum number of flow runs to return offset: Number of flow runs to skip state_type: Filter by state type (e.g., "RUNNING", "COMPLETED", "FAILED")
Returns: A list of flow runs for the specified flow
Inputs: flow_id (required), limit, offset, state_type.
get_flowsB
Get a list of flows with optional filtering.
Args: limit: Maximum number of flows to return offset: Number of flows to skip flow_name: Filter flows by name tags: Filter flows by tags created_after: ISO formatted datetime string for filtering flows created after this time created_before: ISO formatted datetime string for filtering flows created before this time
Returns: A list of flows with their details
Inputs: created_after, created_before, flow_name, limit, +2 more.
get_healthC
Get health status of the Prefect server.
Returns: Health status information
Inputs: no named parameters.
get_task_runC
Get details of a specific task run by ID.
Args: task_run_id: The task run UUID
Returns: Task run details
Inputs: task_run_id (required).
get_task_runsB
Get a list of task runs with optional filtering.
Args: limit: Maximum number of task runs to return offset: Number of task runs to skip task_name: Filter by task name state_type: Filter by state type (e.g., "RUNNING", "COMPLETED", "FAILED") state_name: Filter by state name tags: Filter by tags start_time_before: ISO formatted datetime string start_time_after: ISO formatted datetime string
Returns: A list of task runs with their details
Inputs: limit, offset, start_time_after, start_time_before, +4 more.
get_task_runs_by_flow_runB
Get task runs for a specific flow run.
Args: flow_run_id: The flow run UUID limit: Maximum number of task runs to return offset: Number of task runs to skip state_type: Filter by state type (e.g., "RUNNING", "COMPLETED", "FAILED")
Returns: A list of task runs for the specified flow run
Inputs: flow_run_id (required), limit, offset, state_type.
get_variableC
Get a variable by name.
Args: name: The variable name
Returns: Variable details
Inputs: name (required).
get_variablesB
Get a list of variables with optional filtering.
Args: limit: Maximum number of variables to return offset: Number of variables to skip name: Filter by name pattern
Returns: A list of variables with their details
Inputs: limit, name, offset.
get_work_queueC
Get details of a specific work queue by ID.
Args: work_queue_id: The work queue UUID
Returns: Work queue details
Inputs: work_queue_id (required).
get_work_queue_by_nameC
Get a work queue by name.
Args: name: The work queue name
Returns: Work queue details
Inputs: name (required).
get_work_queuesC
Get a list of work queues with optional filtering.
Args: limit: Maximum number of work queues to return offset: Number of work queues to skip name: Filter by name is_paused: Filter by paused status
Returns: A list of work queues with their details
Inputs: is_paused, limit, name, offset.
get_workspaceC
Get a workspace by ID.
Args: workspace_id: The workspace UUID
Returns: Workspace details
Inputs: workspace_id (required).
get_workspace_by_handleC
Get a workspace by its handle.
Args: account_handle: The account handle workspace_handle: The workspace handle
Returns: Workspace details
Inputs: account_handle (required), workspace_handle (required).
get_workspacesB
Get a list of accessible workspaces.
Args: limit: Maximum number of workspaces to return offset: Number of workspaces to skip name: Filter workspaces by name (note: filtering may not be supported by all Prefect versions)
Returns: A list of workspaces with their details
Inputs: limit, name, offset.
pause_deployment_scheduleC
Pause a deployment's schedule.
Args: deployment_id: The deployment UUID
Returns: Confirmation message
Inputs: deployment_id (required).
pause_work_queueB
Pause a work queue.
Args: work_queue_id: The work queue UUID
Returns: Details of the updated work queue
Inputs: work_queue_id (required).
restart_flow_runC
Restart a flow run.
Args: flow_run_id: The flow run UUID
Returns: Details of the new flow run
Inputs: flow_run_id (required).
resume_deployment_scheduleC
Resume a deployment's schedule.
Args: deployment_id: The deployment UUID
Returns: Confirmation message
Inputs: deployment_id (required).
resume_work_queueB
Resume a work queue.
Args: work_queue_id: The work queue UUID
Returns: Details of the updated work queue
Inputs: work_queue_id (required).
set_deployment_scheduleB
Set a deployment's schedule.
Args: deployment_id: The deployment UUID cron: Cron expression for the schedule interval_seconds: Alternative to cron - interval in seconds anchor_date: Required for interval schedules - the anchor date timezone: Timezone for the schedule
Returns: Updated schedule details
Inputs: deployment_id (required), anchor_date, cron, interval_seconds, +1 more.
set_flow_run_stateB
Set a flow run's state.
Args: flow_run_id: The flow run UUID state: The new state to set (e.g., "SCHEDULED", "RUNNING", "COMPLETED", "FAILED") message: An optional message explaining the state change
Returns: Result of the state change operation
Inputs: flow_run_id (required), state (required), message.
set_task_run_stateB
Set a task run's state.
Args: task_run_id: The task run UUID state: The new state to set (e.g., "SCHEDULED", "RUNNING", "COMPLETED", "FAILED") message: An optional message explaining the state change
Returns: Result of the state change operation
Inputs: state (required), task_run_id (required), message.
update_deploymentB
Update a deployment.
Args: deployment_id: The deployment UUID name: New name for the deployment description: New description version: New version tags: New tags parameters: New parameters work_queue_name: New work queue name
Returns: Details of the updated deployment
Inputs: deployment_id (required), description, name, parameters, +3 more.
update_variableC
Update a variable.
Args: name: The variable name value: New value tags: New tags
Returns: Details of the updated variable
Inputs: name (required), tags, value.
update_work_queueB
Update a work queue.
Args: work_queue_id: The work queue UUID name: New name description: New description is_paused: New paused status concurrency_limit: New concurrency limit
Returns: Details of the updated work queue
Inputs: work_queue_id (required), concurrency_limit, description, is_paused, +1 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelogโฆ
TDQS
Scored across 46 tools
Each tool has a clearly distinct purpose targeting specific resources and actions in the Prefect domain, such as flows, deployments, flow runs, task runs, variables, work queues, blocks, and workspaces. There is no ambiguity or overlap; for example, get_flow_runs and get_task_runs are clearly separated by resource type, and create/update/delete operations are consistently differentiated.
Tool names follow a highly consistent verb_noun pattern throughout, with verbs like get, create, update, delete, pause, resume, set, and cancel paired with specific nouns. All names use snake_case uniformly, and patterns like get_X and get_Xs for list operations are predictable, making the set easy to navigate.
With 46 tools, the count is high and may feel heavy for typical MCP server purposes, though it aligns with Prefect's comprehensive API coverage. While each tool serves a distinct function, the sheer number could overwhelm agents, placing it in the borderline range where 16-25 tools is considered heavy, and 46 exceeds that significantly.
The tool set provides complete CRUD/lifecycle coverage for all core Prefect resources, including flows, deployments, flow runs, task runs, variables, work queues, blocks, and workspaces. It supports operations like creation, retrieval, updating, deletion, state management, scheduling, and health checks, with no obvious gaps that would hinder agent workflows.
Maintenance
Related MCP Connectors
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yoโฆ
A Model Context Protocol server for Wix AI tools
Model Context Protocol server for the Apideck Unified API. Connect any MCP-compatible agent framework to 100+ accounting systems, HRIS platforms, file storage providers, and more through one integration. More information https://www.apideck.com/mcp-server
Nifty's MCP server โ exposes tasks, projects, messages, and files as tools for AI agents.
Related MCP Servers
- AlicenseNot gradedqualityAmaintenanceA Model Context Protocol server that enables AI assistants to interact with Coolify instances through natural language, allowing management of servers, applications, databases, and deployments.6,409 npm625MIT
- AlicenseNot gradedqualityDmaintenanceA Model Context Protocol server that enables AI assistants to interact with n8n workflows through natural language, supporting actions like listing, creating, updating, executing and monitoring workflows.246 npm1,631MIT
- AlicenseAqualityDmaintenanceA server that enables interacting with Prefect workflow automation tools through the Model Context Protocol, offering enhanced reliability through uvx running mechanism and seamless integration with Cursor IDE.177MIT

prefect-mcp-serverofficial
AlicenseAqualityCmaintenanceAn MCP server for interacting with Prefect resources, enabling AI assistants to monitor, manage, and debug workflows.1455MIT