generated: '2026-07-20'
method: searched
status: published
source: https://www.nixtla.io/blog/genai-announcement
server:
  name: nixtla
  transport: stdio
  description: Nixtla ships MCP tools that expose TimeGPT time-series forecasting and anomaly detection
    to AI agents. Announced 2025-12-18 as part of Nixtla Enterprise's foundation-models + MCP + agentic
    release. Distributed via installers for production and local environments; documented to work with
    Claude Desktop, Claude Code, and Cursor. No public hosted/remote endpoint URL is advertised — installed
    locally against the Nixtla API key.
  clients:
  - Claude Desktop
  - Claude Code
  - Cursor
notes: The tools[] below are candidate tools DERIVED from the published OpenAPI operations (https://nixtla.io/docs/openapi.json),
  not a verbatim capture of Nixtla's shipped MCP tool manifest, which is not publicly enumerated. They
  reflect the API surface an MCP server over this API would expose.
tools:
- name: forecast
  description: Multi-series time-series forecast with TimeGPT.
  source_operation: openapi/nixtla-openapi-original.json#v2_forecast_v2_forecast_post
- name: detect_anomalies
  description: Detect anomalies across multiple historical time series.
  source_operation: openapi/nixtla-openapi-original.json#v2_anomaly_detection_v2_anomaly_detection_post
- name: detect_anomalies_online
  description: Real-time (online) anomaly detection over streaming series.
  source_operation: openapi/nixtla-openapi-original.json#v2_online_anomaly_detection_v2_online_anomaly_detection_post
- name: cross_validation
  description: Rolling-window cross validation for multiple series.
  source_operation: openapi/nixtla-openapi-original.json#v2_cross_validation_v2_cross_validation_post
- name: finetune
  description: Fine-tune TimeGPT on custom data and save the model for reuse.
  source_operation: openapi/nixtla-openapi-original.json#v2_finetune_v2_finetune_post
- name: list_finetuned_models
  description: List saved fine-tuned models.
  source_operation: openapi/nixtla-openapi-original.json#v2_finetuned_models_v2_finetuned_models_get
- name: get_finetuned_model
  description: Retrieve metadata for a fine-tuned model.
  source_operation: openapi/nixtla-openapi-original.json#v2_finetuned_model_v2_finetuned_models__finetuned_model_id__get
- name: delete_finetuned_model
  description: Delete a saved fine-tuned model.
  source_operation: openapi/nixtla-openapi-original.json#v2_finetuned_models_delete_v2_finetuned_models__finetuned_model_id__delete
- name: get_model_params
  description: Get model input_size and horizon parameters for a model/frequency.
  source_operation: openapi/nixtla-openapi-original.json#get_model_params_model_params_get
- name: validate_api_key
  description: Validate a Nixtla API key.
  source_operation: openapi/nixtla-openapi-original.json#validate_api_key_validate_api_key_get
deployment:
  mode: local-stdio
  verified: searched
  tools: 10
  checked: '2026-08-12'
  source: catalog MCP census