Fuel Intelligence

Use this for normal natural-language EVChargePoint fuel requests in the ChatGPT app. It always calls fuel_query_resolver first, follows any required recommended next data tool, then returns the correct rendered UI payload. It uses render_fuel_cards for cheapest or selected-fuel station comparison, render_station_fuel_price_matrix for all-fuel station inventory, and render_fuel_table for ranking or insight results. It uses render_fuel_map when the request explicitly says map, maps, map view, in maps, directions, or similar map wording. Prefer this tool over manually chaining lower-level tools unless the user explicitly needs step-by-step tool control.

handle_fuel_query

Use this from the EVChargePoint map widget after the user chooses Search this area. It calls fuel_nearby with exact map coordinates and resolved filters, then returns a refreshed fuel_map payload. This is read-only and must not be used to guess fuel_type_id or brand_id.

refresh_fuel_map

Use this after an EVChargePoint station-list tool returns priced station results. Pass the exact previous tool structuredContent, or the exact previous tool-result wrapper containing it, as source_tool_result. Never pass fuel_query_resolver output, and never reconstruct or manually create station or price data.

render_fuel_cards

Use this after an EVChargePoint station-list or station-price tool returns priced station results with coordinates and the user explicitly asked for a map. Pass the exact previous tool structuredContent, or the exact previous tool-result wrapper containing it, as source_tool_result. Never pass fuel_query_resolver output, and never reconstruct or manually create station, coordinate, or price data.

render_fuel_map

Use this after EVChargePoint fuel_station_prices or ranking/insight tools return price rows. Pass the exact previous tool structuredContent, or the exact previous tool-result wrapper containing it, as source_tool_result. Never pass fuel_query_resolver output, and never reconstruct or manually create rows.

render_fuel_table

Use this after EVChargePoint fuel_address_search, fuel_search, fuel_nearby, or fuel_station_prices returns stations with latest_prices and the user asked for all fuel prices, all fuel type prices, list all prices, or station fuel inventory. Pass the exact previous tool structuredContent, or the exact previous tool-result wrapper containing it, as source_tool_result. Never pass fuel_query_resolver output, and never reconstruct or manually create stations, fuel types, or prices.

render_station_fuel_price_matrix

Use this when the user gives station name, address, suburb, or postcode text. The q argument is required; state and limit are optional. Generic fuel-station searches exclude EV-only charging locations unless the user explicitly asks for EV, electric charging, or chargers. If this returns zero results for a postcode, call the returned recommended_next_tool before answering so nearby postcode-area fuel stations can be checked. This is a data-only tool and does not render the ChatGPT app UI by itself. In this ChatGPT app, after this returns priced station results, the next tool call is mandatory. Use render_fuel_map only when the user explicitly asks to show results on a map, maps, map view, directions, or similar map wording. Otherwise use render_fuel_cards for cheapest or single-fuel station comparisons. If the user asked for all fuel prices, all fuel type prices, list all prices, or station fuel inventory, call render_station_fuel_price_matrix with this exact tool structuredContent as source_tool_result before responding. Do not use a Markdown table or duplicate the full station list when a widget is available.

Use this when the user asks for cheapest fuel stations. Results are sorted by ascending price. All filters are optional; use fuel_query_resolver first for natural fuel names, brand names, suburbs, or postcodes. If fuel_type_id is omitted, the endpoint compares each station's cheapest currently available liquid-fuel price. This is a data-only tool and does not render the ChatGPT app UI by itself. In this ChatGPT app, after this returns priced station results, the next tool call is mandatory. Use render_fuel_map only when the user explicitly asks to show results on a map, maps, map view, directions, or similar map wording. Otherwise use render_fuel_cards for cheapest or single-fuel station comparisons. If the user asked for all fuel prices, all fuel type prices, list all prices, or station fuel inventory, call render_station_fuel_price_matrix with this exact tool structuredContent as source_tool_result before responding. Do not use a Markdown table or duplicate the full station list when a widget is available.

Use this only when exact latitude and longitude are already known, such as from browser location or a geocoded coordinate. A postcode alone is not an exact coordinate; call fuel_query_resolver for postcode/suburb radius phrasing, then use fuel_nearby when the resolver returns a lat/lng recommended_next_tool. This is a data-only tool and does not render the ChatGPT app UI by itself. In this ChatGPT app, after this returns priced station results, the next tool call is mandatory. Use render_fuel_map only when the user explicitly asks to show results on a map, maps, map view, directions, or similar map wording. Otherwise use render_fuel_cards for cheapest or single-fuel station comparisons. If the user asked for all fuel prices, all fuel type prices, list all prices, or station fuel inventory, call render_station_fuel_price_matrix with this exact tool structuredContent as source_tool_result before responding. Do not use a Markdown table or duplicate the full station list when a widget is available.

Resolve a user's natural-language fuel question into state, postcode, fuel type, brand, station candidates, and approximate nearby matches. Call this before search/price tools when the user uses natural names or radius phrasing instead of internal ids or coordinates. For postcode/suburb radius cheapest queries, do not answer from resolver station_matches alone; call the returned recommended_next_tool when it is fuel_nearby to get the radius candidate set. Always use this first for natural-language fuel, brand, state, suburb, postcode, station, or radius terms. If the result includes recommended_next_tool, this tool result is not a final answer. Call recommended_next_tool.name with recommended_next_tool.arguments exactly before responding. Never render cards or tables from resolver output.