PromptLayer MCP server

What the model can do

PromptLayer prompt registry and LLM request logs: prompt templates with their versions, release labels and message text, agent workflows, request-log search and analytics as figures (cost, tokens, latency, models, prompts, errors, never the prompts' inputs or outputs), evaluations and tables; when you ask, can publish a new prompt version or a new prompt, create, move or delete a release label, score or tag a request, create a folder, run an agent workflow or an evaluation (both spend LLM credits), rename or archive an evaluation, add rows to a table and delete a table.

Reads

• promptlayer_list_prompt_templatesPromptLayer prompt templates in the registry: id, name, latest version, chat or completion, model and provider, commit message and tags…
• promptlayer_get_prompt_templateOne PromptLayer prompt template: its messages (role and text) or completion text, input variables, model, commit message…
• promptlayer_list_workflowsPromptLayer agent workflows: id, name, latest version and release labels.
• promptlayer_search_requestsLogged LLM requests in PromptLayer, as figures: id, prompt name, model, provider, input and output tokens, cost, start and end time, latency…
• promptlayer_request_analyticsTotals over logged LLM requests in PromptLayer: request count, cost, tokens, cached and thinking tokens, average and percentile latency…
• promptlayer_get_requestOne logged LLM request's figures: provider, model, tokens, cost, start and end time, latency and trace id.
• promptlayer_list_evaluationsPromptLayer evaluations (evaluation pipelines, also called reports): id, name, comment, dataset, created and updated…
• promptlayer_list_tablesPromptLayer tables (datasets for evaluation): id, title, folder, number of sheets, created and updated.

Acts, when you ask

These change something in PromptLayer. 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.

• promptlayer_publish_prompt_versionPublishes a new version of an existing PromptLayer prompt, built from its latest version…
• promptlayer_create_prompt_templateCreates a new PromptLayer prompt template with its first version: a chat prompt from messages or a completion prompt from text…
• promptlayer_create_release_labelCreates a release label (e.g. "staging") on one version of a PromptLayer prompt.
• promptlayer_move_release_labelMoves a PromptLayer release label to another version of its prompt.
• promptlayer_delete_release_labelDeletes a PromptLayer release label.
• promptlayer_score_requestRecords a score from 0 to 100 on one logged PromptLayer request, under a score name (default "default").
• promptlayer_add_request_metadataAdds key-value metadata to one logged PromptLayer request (for example a customer tier or an experiment name), for filtering later.
• promptlayer_create_folderCreates a folder in the PromptLayer workspace, at the root or inside parent_id.
• promptlayer_run_workflowRuns a PromptLayer agent workflow with the given input variables.
• promptlayer_run_evaluationRuns a PromptLayer evaluation pipeline over its dataset (or the dataset given), creating a new report under the name you give.
• promptlayer_rename_evaluationRenames a PromptLayer evaluation pipeline and optionally sets its tags.
• promptlayer_archive_evaluationArchives (deletes) a PromptLayer evaluation pipeline, removing it from the workspace's active list.
• promptlayer_add_table_rowsAppends rows to one sheet of a PromptLayer table.
• promptlayer_delete_tableDeletes a PromptLayer table (a soft delete: it leaves the workspace's active lists).

Connect it

First connect PromptLayer on Dashboard → Connections, then add Memtro to your client once. The PromptLayer 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 PromptLayer 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 did our LLM requests cost last week?
• › Which version of the support prompt is labelled prod?
• › Which prompt had the most errors yesterday?
• › What is the p95 latency for gpt-4o this month?
• › When did the onboarding evaluation last run?

Questions

Is this an official PromptLayer MCP server?

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

Only when you ask, and only what it has been granted: publish a new version of a prompt (replace its messages or text, optionally move release labels to it); create a new prompt template; create a release label on a prompt version; move a release label to another version (changes what code fetching that label gets); delete a release label; score a logged request (0 to 100); add metadata to a logged request; create a folder; run an agent workflow (spends LLM provider credits); run an evaluation (spends LLM provider credits); rename an evaluation; archive an evaluation; add rows to a table sheet; delete a table. 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