MCP App Store
by KeenEthics
PredictHQ icon

PredictHQ MCP Server for Claude and ChatGPT

By PredictHQ

Overview

Ask about any location and date range, and Claude answers from verified real-world context: the concerts, conferences, sports, and holidays that move demand. Search and rank events, quantify their impact as model-ready features, get event-driven forecasts, and trace any demand shift to the specific events behind it - the context behind pricing, staffing, and inventory decisions. This is Grounding with PredictHQ: your AI answering from facts, not guesses.

How to connect

  1. Claude: Open the PredictHQ listing in the Claude connectors directory and choose Connect. Claude listing PredictHQ exposes 58 tools here.
  2. ChatGPT: Open the PredictHQ listing in the ChatGPT app directory and choose Connect. ChatGPT listing PredictHQ exposes 58 tools here.

Tools

auth_get_subscription

ChatGPT
Get the authenticated user's subscription details, including visibility window and covered places.

beam_api_create_analysis

ChatGPT
Creating an Analysis is the first step in understanding which types of events impact your demand.

beam_api_create_analysis_group

ChatGPT
Creating an Analysis Group allows to aggregate Feature Importance for a number of Analyses

beam_api_delete_analysis

ChatGPT
Delete an existing Analysis. The Analysis will no longer appear in default search results, but can still be retrieved by filtering with status=deleted. Any Analysis Groups containing this Analysis will be automatically refreshed.

beam_api_delete_analysis_group

ChatGPT
Delete an existing Analysis Group. The individual Analyses within the group are not deleted.

beam_api_get_analysis

ChatGPT
Returns the full details of a single Analysis, including its location, demand type, rank, readiness status, and processing state.

beam_api_get_analysis_correlation

ChatGPT
Retrieve results of the correlation of our event data with your transactional demand data.

beam_api_get_analysis_feature_importance

ChatGPT
This endpoint provides the relevant ML features (based on feature importance testing) that are shown to impact the demand for the Beam Analysis. These are the features you should take into your forecasting model to improve your model accuracy. These values represent each group of features' statistical significance when it comes to impacting observable incremental/decremental changes in demand. The easiest way to get these ML features from our Features API to be used in your models is by using the Beam analysis_id in your Features API request.

beam_api_get_analysis_group

ChatGPT
Returns the full details of a single Analysis Group, including its member analyses, demand type, readiness status, and processing state.

beam_api_get_analysis_group_feature_importance

ChatGPT
This endpoint provides Feature Importance results aggregated across the Analyses in an existing Analysis Group, and returns a list of feature groups with associated Features API features and group p-values.

beam_api_list_analyses

ChatGPT
Returns a paginated list of your Beam Analyses. Supports filtering by status, readiness_status, industry, interval, label, and date range. Note: Polygon and MultiPolygon geometry is omitted from results to reduce token usage. To retrieve full geometry for a specific analysis, call the get_analysis tool.

beam_api_list_analysis_groups

ChatGPT
Returns a paginated list of your Analysis Groups. Supports filtering by status, readiness_status, industry, interval, and date range.

beam_api_patch_analysis

ChatGPT
Update one or more fields on an existing Analysis without replacing the entire record. Only the fields you provide will be updated; all other fields remain unchanged.

beam_api_patch_analysis_group

ChatGPT
Partially update an existing Analysis Group. This endpoint accepts the same request body fields as the Create an Analysis Group endpoint. Remember this is a PATCH endpoint which means only the fields you provide will be updated. If you wish to replace all fields please use the PUT endpoint.

beam_api_refresh_analysis

ChatGPT
Rerun the Beam correlation and analysis process for an existing Analysis. The readiness_status will be set to pending until processing completes. Returns 400 if the Analysis is already being processed.

beam_api_refresh_analysis_group

ChatGPT
Rerun the Beam Analysis Group aggregation process. The readiness_status will be set to pending until processing completes. Returns 400 if the Group is already being processed.

beam_api_update_analysis

ChatGPT
Update (replace) an existing Analysis. This endpoint accepts the same request body fields as the Create an Analysis endpoint. Remember this is a PUT endpoint which means you must provide all supported fields or you may lose data - you are effectively replacing the existing record with a new record containing all the fields you provide. We recommend first getting the existing record and pre-populating the request body with the current values, then change the fields you need to change. If you wish to update a smaller number of fields please use the PATCH endpoint.

beam_api_update_analysis_group

ChatGPT
Update (replace) an existing Analysis Group. This endpoint accepts the same request body fields as the Create an Analysis Group endpoint. Remember this is a PUT endpoint which means you must provide all supported fields - you are effectively replacing the existing record with a new record containing all the fields you provide. We recommend first getting the existing record and pre-populating the request body with the current values, then change the fields you need to change.

beam_api_upload_analysis_demand

ChatGPT
Upload historical demand data to a Beam analysis. This is required before Beam can run correlation and feature importance. Daily data requires at least 6 months (180 data points). Weekly data requires at least 2 years (104 data points), with all dates on the same weekday. Uploading new data triggers reprocessing. Existing data for the same dates is overwritten. All demand values must be positive numbers. Provide demand as a CSV string via demand_data with header row date,demand followed by data rows.

broadcasts_api_count_broadcasts

ChatGPT
Returns the total number of broadcast records matching the given filters. Accepts the same filter parameters as the search endpoint but returns only a count, without pagination or result data.

broadcasts_api_list_broadcasts

ChatGPT
Returns a paginated list of broadcast records matching the given filters. Broadcasts represent scheduled or predicted Live TV events at the county level in the US. Use date range filters (start.*, updated.*, first_seen.*) to narrow results by time, location.place_id or location.origin to filter by geography, and event.* filters to find broadcasts linked to specific physical events. Results are sorted by start ascending by default. Use the sort parameter to change ordering. Pagination is offset-based.

docs_get_page

ChatGPT
Get a specific page from PredictHQ's technical documentation. Use this tool to retrieve the full content of a specific documentation page when you already know the exact URL. Always obtain URLs from docs_search results — do NOT guess or construct URLs from patterns, as documentation page paths do not follow a predictable naming convention and guessed URLs will return "Page not found" errors. Correct workflow: 1. Call docs_search to find relevant pages and get their exact URLs. 2. Call docs_get_page with a URL returned by docs_search.

events_api_count_events

ChatGPT
Get the count of events by category, PHQ Label and more. The Count endpoint supports the same query parameters as the Search Events endpoint.

events_api_get_event

ChatGPT
Get a single event by ID with full details including geometry. Use this when you need the complete event record for a specific event — including full Polygon/MultiPolygon geometry that is omitted from list_events results. Prefer list_events for discovery and get_event for detail retrieval.

events_api_list_events

ChatGPT
Search for events happening in a location and date range. Use our extensive filters to narrow down your results. Note: Polygon and MultiPolygon geometry is omitted from results to reduce token usage. To retrieve full geometry for a specific event, call the get_event tool.

features_api_get_features

ChatGPT
Returns aggregated ML-ready features for the specified location and date range. Returns pre-built ML-ready features for the specified location and date range, aggregated per day or per week. Features are opt-in: only features included in the request body are returned. Beam (recommended): Provide a beam object with an analysis_id to automatically derive location, interval, and feature selection from a Beam analysis. This is the recommended approach. When beam.analysis_id is provided, you must not specify location, interval, or individual features separately — these are determined by the analysis and the request will be rejected if they are included. Manual configuration: If not using Beam, location must be specified via location.geo (lat/lon + radius), location.place_id, or location.saved_location_id, and features must be explicitly requested. Feature families: phq_attendance_*, phq_impact_*, phq_spend_*, and phq_viewership_* features return statistical summaries (count, sum, min, max, avg, median, std_dev) — pass true for defaults (count and sum), or a criteria object to customise stats and filter by phq_rank or local_rank. phq_rank_* features return a rank_levels object (counts bucketed into 5 severity levels) and are requested by setting the field to true. Date range (active) is required in all cases and uses local time. Multi-day occurrences are included for each day they are active within the range. Results are paginated with offset and limit, and can be returned as JSON or CSV via the Accept header.

forecasts_api_create_demand_data

ChatGPT
Upload historical demand data to a forecast model. This is required before training the model. Uploading new data for existing dates overwrites previous values. Provide demand as a CSV string via demand_data with header row date,demand followed by data rows.

forecasts_api_create_model

ChatGPT
Creates a new forecast model. You must provide a name and a location (via a Saved Location ID). Optionally specify demand_type, label, external_id, forecast_window, algo, and generate_baseline. After creating a model, upload historical demand data using the Create Demand Data endpoint, then trigger training.

forecasts_api_delete_model

ChatGPT
Permanently deletes a forecast model and all associated demand data and forecasts. The model must not be in preparing, training, or forecasting status. This action is irreversible.

forecasts_api_get_forecast

ChatGPT
Returns the generated forecast results for a trained model. Each result includes a date and predicted demand value. Supports filtering by date range.

forecasts_api_get_model

ChatGPT
Retrieves a single forecast model by ID, including its current readiness status, configuration, accuracy metrics, and location details.

forecasts_api_list_algorithms

ChatGPT
Returns the list of available forecasting algorithms. Currently one algorithm is offered (phq-xgboost) based on XGBoost.

forecasts_api_list_demand_data

ChatGPT
Returns the demand data that has been uploaded to the model. Supports filtering by date range. By setting the Accept header, you can specify the format of the response. The default is JSON. Supported formats: application/json, text/csv, application/x-ndjson.

forecasts_api_list_models

ChatGPT
Returns a paginated list of forecast models belonging to the authenticated organization. Supports full-text search across name, label, and external_id via the q parameter, and filtering by Saved Location ID and readiness status.

forecasts_api_patch_model

ChatGPT
Partially updates an existing forecast model. Only the fields provided in the request body will be changed. Changes to location, demand_type, forecast_window, algo, or generate_baseline are blocked while the model is in preparing, training, or forecasting status. Changing any of those fields sets the model to stale status, requiring re-training to take effect.

forecasts_api_train_model

ChatGPT
Triggers training for the specified forecast model. The model must have demand data uploaded and be in a state that allows training (e.g. ready or created with demand data). Training runs asynchronously — poll the model's readiness.status field to monitor progress.

forecasts_api_update_model

ChatGPT
Replaces the full configuration of an existing forecast model. This is a full replacement — all mutable fields should be provided. Changes to location, demand_type, forecast_window, algo, or generate_baseline are blocked while the model is in preparing, training, or forecasting status. Changing any of those fields sets the model to stale status, requiring re-training to take effect.

geocode_api_autocomplete_places

ChatGPT
Search for places by name or address using autocomplete. Returns Google Place results, suitable for venues, businesses, street addresses, and ambiguous POIs. The google_place_id returned by this tool is a Google Place ID and is not compatible with list_places(id=…) in the Places server. For administrative-only lookups (a city/region/country with hierarchy info), prefer list_places." Returns a list of matching places with their Google Place ID. Use to convert a place name or address provided by the user into coordinates before calling any API that requires a lat/lon. Examples: 'AT&T Stadium Dallas', '123 Main St Chicago', 'Wembley Stadium London'. Always use this before creating a Saved Location or calling the Predicted Impact Area API when you have a place name but not coordinates. Returns multiple candidates — use the most relevant result based on context, or ask the user to confirm if ambiguous (e.g. 'Springfield' could be many cities).

geocode_api_get_place

ChatGPT
Get full details for a Google Place using a Google Place ID previously retrieved from autocomplete_places. Returns the Google Place formatted address, coordinates, and additional details. Only call this when you have a google_place_id from a prior autocomplete_places call. Do not guess or construct Google Place IDs. Use when you need full place details beyond what autocomplete returned, or when you need to confirm the exact coordinates before creating a Saved Location.

impact_area_api_get_impact_area

ChatGPT
Returns the Predicted Impact Area for a given location and industry as a GeoJSON feature. Two response shapes are supported via area_type: - polygon (default, recommended) — a GeoJSON Polygon or MultiPolygon that precisely follows the true boundaries of the impact area. This is the best practice as real impact areas are rarely circular. - radius — a GeoJSON Point with a radius property defining a circular approximation. Supported but not recommended, as it sacrifices accuracy for simplicity. The returned GeoJSON cannot be passed directly to the Events, Features, or Beam APIs. To query events or features within this area, create a Saved Location using origin_geojson — the Predicted Impact Area will be calculated and stored automatically, and the location_id can then be used across those APIs.

places_api_list_place_hierarchies

ChatGPT
Return the administrative hierarchy chain for one or more places. Exactly one of location.place_id or location.origin must be provided — supplying both or neither returns 400. By place ID (location.place_id): Returns stored hierarchies for the given place IDs. Results are paginated over the matched places. The country filter is not applied in this mode. By coordinates (location.origin): Finds the nearest locality to the given lat,lon and returns its hierarchy chain. - When level is region or above (planet, continent, country, region), only nearby localities are considered. - When level is more granular than region (county, localadmin, locality, neighbourhood) or not specified, the search is extended to include major and metro areas within the same region as the nearest locality. - If no locality is found at the exact coordinates, a 500 km fallback radius is applied before giving up. - Use country (comma-separated ISO 3166-1 alpha-2 codes) to restrict the coordinate lookup to specific countries. Use level to scope which hierarchy levels are returned. Results are paginated with offset and limit.

places_api_list_places

ChatGPT
Search for places by text query, country, geographic coordinates, or place IDs. At least one of q, country, location, or id must be provided — otherwise a 400 is returned. Filtering: - q — free-text search on place name. - country — comma-separated ISO 3166-1 alpha-2 country codes. - location — coordinates in lat,lon or @lat,lon format. Lat must be between −90 and 90, lon between −180 and 180. - id — comma-separated place IDs. When id is the only filter and no pagination parameters are set, an optimised multi-get is used instead of a search query. - type — comma-separated place types. Accepts admin level values as well as aliases (e.g. local expands to neighbourhood, locality, and localadmin). Results are paginated with offset and limit.

saved_locations_api_create_closure

ChatGPT
Creates a closure for a location. A closure represents a period during which the location is closed. The effective_from and effective_to date range must not overlap with existing closures for the same location.

saved_locations_api_create_opening_hours

ChatGPT
Creates a new opening hours record for a location. Slots define the open/close times for each day of the week. The effective_from and effective_to date range must not overlap with existing opening hours for the same location.

saved_locations_api_create_saved_location

ChatGPT
Create a new Saved Location. The recommended approach is to provide only origin_geojson (a GeoJSON Point at the exact location of the business) together with an industry value, without specifying geojson. PredictHQ will automatically calculate and store a Predicted Impact Area in the geojson field, calibrated for the business type. Only specify geojson directly if you need a custom area rather than the auto-calculated Predicted Impact Area. Use place_ids instead of geojson when the location represents an entire city, state, or country. Returns a location_id that can be reused across the Events, Features, and Beam APIs without re-supplying coordinates.

saved_locations_api_delete_closure

ChatGPT
Deletes a closure by ID for the specified location.

saved_locations_api_delete_opening_hours

ChatGPT
Deletes an opening hours record by ID for the specified location.

saved_locations_api_delete_saved_location

ChatGPT
Permanently delete a saved location by ID. This action is irreversible — the location and all associated insights data will be removed.

saved_locations_api_get_closure

ChatGPT
Returns a single closure by ID for the specified location.

saved_locations_api_get_opening_hours

ChatGPT
Returns a single opening hours record by ID for the specified location.

saved_locations_api_get_saved_location

ChatGPT
Retrieve full details for a single Saved Location by ID. Response includes the geojson field (which stores the Predicted Impact Area boundary), origin_geojson (exact business point), industry, labels, formatted_address, and summary_insights — which contain attended/non-attended event counts, PHQ attendance sum, and Predicted Event Spend metrics for the next 90-day window. The status field reflects enrichment state: pending means the Predicted Impact Area and summary_insights are still being calculated (typically resolves within seconds); active means the location is fully enriched and ready to use.

saved_locations_api_list_closures

ChatGPT
Returns a paginated list of closures for a location. Optionally filter by date range using active_from and active_to.

saved_locations_api_list_opening_hours

ChatGPT
Returns a paginated list of opening hours records for a location. Optionally filter by date range using active_from and active_to.

saved_locations_api_list_saved_location_insight_events

ChatGPT
Returns events that impact the specified Saved Location. Please refer to the Events API docs for the available query parameters in addition to those listed below.

saved_locations_api_list_saved_locations

ChatGPT
Search and list Saved Locations. Supports full-text search across name, description, location_code, formatted_address, and labels via the q parameter. Saved Locations are standalone geographic entities that exist independently of any other PredictHQ product. Creating a Saved Location does not create a Beam Analysis — those must be created separately. The returned location_id values can be passed directly to the Events API, Features API, and Beam API without needing to supply coordinates each time. Note: Polygon and MultiPolygon geometry is omitted from results to reduce token usage. To retrieve full geometry for a specific saved location, call the get_saved_location tool.

saved_locations_api_update_opening_hours

ChatGPT
Replaces an existing opening hours record. The effective_from and effective_to date range must not overlap with other opening hours for the same location.

saved_locations_api_update_saved_location

ChatGPT
Fully replace an existing Saved Location. This is a PUT (full replacement) — all fields should be provided, not just the ones being changed. To trigger a recalculation of the Predicted Impact Area, provide origin_geojson without specifying geojson, the same as when creating a location.

auth_get_subscription

Claude

beam_api_create_analysis

Claude

beam_api_create_analysis_group

Claude

beam_api_delete_analysis

Claude

beam_api_delete_analysis_group

Claude

beam_api_get_analysis

Claude

beam_api_get_analysis_correlation

Claude

beam_api_get_analysis_feature_importance

Claude

beam_api_get_analysis_group

Claude

beam_api_get_analysis_group_feature_importance

Claude

beam_api_list_analyses

Claude

beam_api_list_analysis_groups

Claude

beam_api_patch_analysis

Claude

beam_api_patch_analysis_group

Claude

beam_api_refresh_analysis

Claude

beam_api_refresh_analysis_group

Claude

beam_api_update_analysis

Claude

beam_api_update_analysis_group

Claude

beam_api_upload_analysis_demand

Claude

broadcasts_api_count_broadcasts

Claude

broadcasts_api_list_broadcasts

Claude

docs_get_page

Claude

events_api_count_events

Claude

events_api_get_event

Claude

events_api_list_events

Claude

features_api_get_features

Claude

forecasts_api_create_demand_data

Claude

forecasts_api_create_model

Claude

forecasts_api_delete_model

Claude

forecasts_api_get_forecast

Claude

forecasts_api_get_model

Claude

forecasts_api_list_algorithms

Claude

forecasts_api_list_demand_data

Claude

forecasts_api_list_models

Claude

forecasts_api_patch_model

Claude

forecasts_api_train_model

Claude

forecasts_api_update_model

Claude

geocode_api_autocomplete_places

Claude

geocode_api_get_place

Claude

impact_area_api_get_impact_area

Claude

places_api_list_place_hierarchies

Claude

places_api_list_places

Claude

saved_locations_api_create_closure

Claude

saved_locations_api_create_opening_hours

Claude

saved_locations_api_create_saved_location

Claude

saved_locations_api_delete_closure

Claude

saved_locations_api_delete_opening_hours

Claude

saved_locations_api_delete_saved_location

Claude

saved_locations_api_get_closure

Claude

saved_locations_api_get_opening_hours

Claude

saved_locations_api_get_saved_location

Claude

saved_locations_api_list_closures

Claude

saved_locations_api_list_opening_hours

Claude

saved_locations_api_list_saved_location_insight_events

Claude

saved_locations_api_list_saved_locations

Claude

saved_locations_api_update_opening_hours

Claude

saved_locations_api_update_saved_location

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/predicthq"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/predicthq)