Dagster+ MCP server

What the model can do

Dagster+ data orchestration: runs by job, status and date with step counts, code locations and their jobs, schedules and sensors and whether they are on, assets with their last materialization, and backfills; when you ask, can launch, re-execute, terminate or delete runs, start or stop schedules and sensors, reload a code location, cancel a backfill, and report an external asset materialization or check result.

Reads

• dagster_runsDagster+ runs, newest first: run id, job, status, created, start and end times, and the partition, backfill, schedule or sensor that started it.
• dagster_runOne Dagster+ run in detail: status, times, steps succeeded and failed, materializations, the assets it targeted…
• dagster_jobsDagster+ code locations and repositories with their jobs.
• dagster_automationsDagster+ schedules (cron, timezone, job) and sensors (type, minimum interval) across code locations, and whether each is running or stopped.
• dagster_assetsDagster+ assets: key, group, description, compute kind, jobs that materialize it, and when it was last materialized.
• dagster_backfillsDagster+ backfills, newest first: id, status, title, number of partitions, the job or assets targeted, and when created and ended.

Acts, when you ask

These change something in Dagster+. 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.

• dagster_launch_runLaunches a run of a Dagster+ job, which executes the pipeline's code against the company's systems…
• dagster_reexecute_runRe-executes a Dagster+ run as a new run: from the failed steps (FROM_FAILURE, the default), from failed assets (FROM_ASSET_FAILURE), or all steps…
• dagster_terminate_runTerminates an in-progress Dagster+ run.
• dagster_delete_runPermanently deletes a Dagster+ run's record and its event log from the deployment; what the run did in other systems is untouched.
• dagster_start_scheduleStarts (turns on) a Dagster+ schedule, so it launches its job on its cron from now on, running pipeline code and spending compute each time.
• dagster_stop_scheduleStops (turns off) a Dagster+ schedule, so its job is no longer launched on its cron until it is started again.
• dagster_start_sensorStarts (turns on) a Dagster+ sensor, so it evaluates on its interval and launches runs when its condition is met.
• dagster_stop_sensorStops (turns off) a Dagster+ sensor, so it no longer launches runs until started again.
• dagster_reload_code_locationReloads a Dagster+ code location, so its definitions (jobs, schedules, sensors, assets) are read again from its current code…
• dagster_cancel_backfillCancels a Dagster+ backfill: no more of its partitions are launched, and its in-progress runs are canceled.
• dagster_report_materializationRecords in Dagster+ that an external asset was materialized (updated) outside Dagster, with optional metadata, data version…
• dagster_report_asset_checkRecords in Dagster+ the result of an asset check run outside Dagster: passed or failed, with severity ERROR (default) or WARN and optional metadata.

Connect it

First connect Dagster+ on Dashboard → Connections, then add Memtro to your client once. The Dagster+ 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 Dagster+ 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

• › Which Dagster runs failed overnight?
• › When was the orders_daily asset last materialized?
• › Which schedules are stopped right now?
• › How far has the March backfill got?
• › Re-execute last night's failed ingest run from failure.

Questions

Is this an official Dagster+ MCP server?

No. It is Memtro's MCP server with Memtro's Dagster+ connector behind it: the connector reads Dagster+ through its API with the credentials you connect on Dashboard → Connections, and Dagster+ 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 Dagster+?

Only when you ask, and only what it has been granted: launch a run of a job; re-execute a run (from failure or all steps); terminate a run; delete a run's record; start a schedule; stop a schedule; start a sensor; stop a sensor; reload a code location; cancel a backfill; report an external asset materialization; report an external asset check result. 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