LangWatch MCP server

What the model can do

LangWatch LLM observability for a project: traces of the AI features with cost, latency, tokens, labels, evaluation verdicts and thumbs feedback (searchable with the Trace Explorer filter language), reviewers' annotations, managed prompts and their versions, online-evaluation monitors, automations (alerts and reports), experiments and their runs, agent simulation runs and datasets; when you ask, can annotate a trace, edit or delete an annotation, create, edit, restore or delete a prompt, switch a monitor or automation on or off, run an experiment (spends model tokens), add dataset records and set a trace's metadata.

Reads

• langwatch_search_tracesSearch the project's LangWatch traces (one per run of an AI feature) in a time window: per trace its id, when it ran and was last updated, labels…
• langwatch_list_annotationsAnnotations reviewers left on LangWatch traces: thumbs up or down, comment and scores, with the trace id.
• langwatch_list_promptsManaged prompts in LangWatch: handle, name, current version, model and settings.
• langwatch_list_monitorsOnline-evaluation monitors (an evaluator scoring live traces or threads): name, evaluator, whether enabled, when it runs…
• langwatch_list_automationsAutomations on the project (alerts, scheduled reports and trace automations), newest first, paused ones included: name, kind, what it does (email…
• langwatch_list_experimentsExperiments (batch evaluations over a dataset): slug, name, type, how many runs and when last run.
• langwatch_list_simulation_runsAgent simulation runs (scenario tests): scenario, batch, status, verdict, which criteria were met or unmet, duration, cost and link.
• langwatch_list_datasetsDatasets in the project (for experiments and evaluations): name, slug, columns, record count and link.

Acts, when you ask

These change something in LangWatch. 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.

• langwatch_annotate_traceAdds an annotation to a LangWatch trace (a comment and thumbs up or down), unattributed, visible to everyone on the project.
• langwatch_update_annotationReplaces an annotation's comment and thumbs on LangWatch.
• langwatch_delete_annotationPermanently deletes an annotation from a LangWatch trace.
• langwatch_create_promptCreates a managed prompt in LangWatch with its first version.
• langwatch_update_promptChanges a LangWatch prompt by saving a new version (the old ones stay and can be restored).
• langwatch_restore_prompt_versionRestores a LangWatch prompt to an earlier version by saving a new version with that version's settings…
• langwatch_delete_promptPermanently deletes a LangWatch prompt and its versions; apps that fetch it by handle stop getting it.
• langwatch_set_monitor_enabledSwitches a LangWatch online-evaluation monitor on or off.
• langwatch_set_automation_enabledResumes (enabled true) or pauses (false) a LangWatch automation.
• langwatch_run_experimentStarts a run of a saved LangWatch experiment over its dataset.
• langwatch_add_dataset_recordsAdds records to a LangWatch dataset, each an object of column name to value (columns from langwatch_list_datasets).
• langwatch_update_trace_metadataSets metadata on a LangWatch trace after the fact: new keys are added and existing ones changed, others kept, and labels, when passed…

Connect it

First connect LangWatch on Dashboard → Connections, then add Memtro to your client once. The LangWatch 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 LangWatch 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 traces failed evaluation yesterday, and what did they cost?
• › What annotations did reviewers leave this week?
• › What changed in the support reply prompt across its versions?
• › Which evaluation monitors are switched off?
• › How did the last run of the summarisation experiment do?

Questions

Is this an official LangWatch MCP server?

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

Only when you ask, and only what it has been granted: annotate a trace with a comment and thumbs up or down; edit an annotation; delete an annotation; create a managed prompt; change a prompt, saving a new version that apps fetching it receive; restore a prompt to an earlier version; delete a prompt; switch an online-evaluation monitor on or off; pause or resume an automation (alerts and reports); run an experiment (calls models, spends tokens); add records to a dataset; set metadata or labels on a trace. 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