Prefect MCP server

What the model can do

Prefect Cloud or a self-hosted Prefect 3 server: flows, flow runs with their state and logs, deployments and their schedules, work pools and automations; and, when asked, run, update, pause, resume or delete a deployment, add, change or remove its schedules, retry, resume, cancel, rename or delete a flow run, delete a flow, pause or create a work pool or queue, create a concurrency limit or markdown artifact, turn an automation on or off, or create, update or delete a variable.

Reads

• prefect_list_flow_runsPrefect flow runs with flow name, state (and its message), expected and actual start, end and run time.
• prefect_get_flow_runOne Prefect flow run with its state and, by default, its most recent log lines (oldest first, each cut short).
• prefect_list_deploymentsPrefect deployments: name, flow, schedules, whether paused, status, work pool and queue, version, entrypoint and when a worker last polled.
• prefect_list_flowsPrefect flows (the code that deployments run): name, tags, created and updated.
• prefect_list_work_poolsPrefect work pools: name, type, status (ready, not ready, paused), whether paused, and concurrency limit.
• prefect_list_deployment_schedulesThe schedules of one Prefect deployment with their ids, cron, interval or rrule, time zone and whether active.
• prefect_list_automationsPrefect automations: id, name, whether enabled, tags and the kinds of action each takes (e.g. run-deployment, send-notification).

Acts, when you ask

These change something in Prefect. The model is told to use them only when you ask for exactly that, and each one is granted by name: per team on a shared connection, by the owner on a personal one.

• prefect_run_deploymentStart a run of a Prefect deployment.
• prefect_cancel_flow_runCancel a Prefect flow run by moving it to CANCELLING; the worker then stops it.
• prefect_create_variableCreate a Prefect variable.
• prefect_update_variableChange a Prefect variable's value and/or tags.
• prefect_delete_variablePermanently delete a Prefect variable by name.
• prefect_pause_deploymentPause a Prefect deployment's schedule.
• prefect_resume_deploymentResume a paused Prefect deployment's schedule.
• prefect_delete_deploymentPermanently delete a Prefect deployment: its schedules stop and it can no longer be run until someone deploys it again from code.
• prefect_delete_flowPermanently delete a Prefect flow record.
• prefect_retry_flow_runRetry a failed or crashed Prefect flow run: sets it back to Scheduled ("AwaitingRetry")…
• prefect_resume_flow_runResume a paused Prefect flow run, optionally giving the input it is waiting for; the flow then CONTINUES executing on the customer's infrastructure.
• prefect_update_flow_runRename a Prefect flow run and/or replace its tags (the tags given replace all the old ones).
• prefect_delete_flow_runPermanently delete a Prefect flow run record with its state history and logs.
• prefect_update_deploymentChange a Prefect deployment's description, version, tags, default parameters, work pool or queue, or concurrency limit.
• prefect_create_deployment_scheduleAdd a schedule to a Prefect deployment (cron, every N seconds, or an RRULE).
• prefect_update_deployment_scheduleChange one schedule of a Prefect deployment: activate or deactivate it, or replace its timing.
• prefect_delete_deployment_schedulePermanently remove one schedule from a Prefect deployment; no more runs are scheduled from it.
• prefect_update_work_poolPause or unpause a Prefect work pool, or change its description or concurrency limit.
• prefect_create_work_poolCreate a Prefect work pool of the given type (e.g. process, docker, kubernetes, or a push type such as cloud-run:push).
• prefect_create_work_queueCreate a work queue inside a Prefect work pool (lower priority number is served first).
• prefect_create_concurrency_limitCreate a Prefect global concurrency limit: flows that take a slot by this name wait while all slots are in use…
• prefect_create_artifactPublish a markdown artifact in Prefect, optionally attached to a flow run, where everyone in the workspace can read it in the UI.
• prefect_set_automation_enabledTurn a Prefect automation on or off.

Connect it

First connect Prefect on Dashboard → Connections, then add Memtro to your client once. The Prefect tools appear alongside the tools of everything else you have connected; nothing is configured per connector.

claude plugin marketplace add memtro-org/claude-memtro
claude plugin install memtro@memtro --config api_key=sk-memtro-...

{
  "mcpServers": {
    "memtro": {
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

{
  "mcp": {
    "memtro": {
      "type": "remote",
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

Who may use it

The key identifies the person, and the server offers that person the tools they may use everywhere else in Memtro. A shared Prefect connection is used by the teams an administrator granted it to, tool by tool, on the Permissions page; a personal connection is used by its owner alone, and its tools that act stay off until the owner allows them. See Teams and permissions and Connectors.

Ask it like this

• › What failed or crashed overnight?
• › Why did the nightly ETL run fail?
• › Which deployments are paused?
• › Why are runs stuck in Pending?
• › Retry last night's failed backfill-orders run

Questions

Is this an official Prefect MCP server?

No. It is Memtro's MCP server with Memtro's Prefect connector behind it: the connector reads Prefect through its API with the credentials you connect on Dashboard → Connections, and Prefect publishes nothing here. The same endpoint carries every other system you connect and Memtro's memory tools.

What does it cost?

Nothing on top of Memtro. One price per organisation by its number of users, from $50 a month for up to 10 users after a 7-day free trial, with every connector, the MCP server, jobs and memory included. You bring your own model keys.

Can it change anything in Prefect?

Only when you ask, and only what it has been granted: run a deployment (schedules a flow run on your infrastructure); cancel a flow run; create a variable; update a variable's value or tags; delete a variable; pause a deployment's schedule (deletes its waiting scheduled runs); resume a deployment's schedule (schedules runs immediately); delete a deployment; delete a flow; retry a failed or crashed flow run (runs it again on your infrastructure); resume a paused flow run; rename or retag a flow run; delete a flow run and its logs; update a deployment's description, version, tags, parameters, work pool or queue, or concurrency limit; add a schedule to a deployment; change, activate or deactivate a deployment schedule; delete a deployment schedule; pause or unpause a work pool, or change its description or concurrency limit; create a work pool; create a work queue in a work pool; create a global concurrency limit; publish a markdown artifact; turn an automation on or off. Every other tool reads, searches or lists. An administrator grants each acting tool by name per team, and a personal connection acts only after its owner allows it.

Which clients work?

Any MCP client that speaks Streamable HTTP: Claude Code, Cursor, Claude Desktop, OpenCode, Codex, Zed, Windsurf, Cline and the Anthropic and OpenAI SDKs. There is no local process to run; the server is remote and authenticated with your key.

Other MCP servers · Engineering and operations

• GitHub MCP server
• AWS CloudWatch MCP server
• Sentry MCP server
• ServiceNow MCP server
• Okta MCP server
• Twilio MCP server