fetch
ChatGPTRetrieve complete document content by ID for detailed analysis and citation. Use this after finding relevant documents with the search tool to get complete information for analysis and proper citation.
generate_report
ChatGPTExecute a Frappe report. IMPORTANT: Always call report_requirements(report_name) FIRST to get mandatory filters and valid options, then call this tool with explicit filters. Missing filters are auto-defaulted (dates, company) which often returns empty data. Supports Script Reports, Query Reports, and Custom Reports. Report Builder reports are not supported. Large/prepared reports are handled automatically with polling.
get_doctype_info
ChatGPTGet DocType metadata and field information
get_document
ChatGPTRetrieve detailed information about a specific Frappe document. Use when users ask for details about a particular record they know the name/ID of.
list_documents
ChatGPTSearch and list Frappe documents with optional filtering. Use this when users want to find records, get lists of documents, or search for data. This is the primary tool for data exploration and discovery.
report_list
ChatGPTDiscover and search available Frappe business reports across all modules. Use this tool to find the appropriate report for business questions before attempting custom data analysis. Returns a list of available reports with descriptions, modules, and types. Available reports include Sales Analytics, Territory Analysis, Customer Reports, P&L Statement, Balance Sheet, Receivables/Payables, Inventory Reports, Item Movement, Stock Valuation, Payroll, Attendance, and Leave Reports. Always search for existing reports first before building custom queries or analysis.
report_requirements
ChatGPTGet report metadata including required and optional filters, columns, and execution requirements for Script Reports, Query Reports, and Custom Reports. Use this tool before executing reports to understand what filters are mandatory, what exact filter values are valid, and how to structure the report request. This prevents filter errors and helps plan successful report execution. Returns complete report metadata including filter definitions with field types (Link, Select, Date), valid enum options for select fields, column structure, report type, and capabilities. IMPORTANT: Use this FIRST before calling generate_report to understand what exact filter values are needed - Link fields require exact database names (e.g., exact Company name, Customer name), Select fields show valid enum values. Essential when generate_report returns filter errors or when planning complex report execution. NOTE: Report Builder reports are not supported as they are simple DocType list views without business logic.
search
ChatGPTSearch for documents using OpenAI Vector Store search. Returns a list of search results with basic information. Use the fetch tool to get complete document content.
search_doctype
ChatGPTSearch within a specific DocType
search_documents
ChatGPTGlobal search across all accessible documents
search_link
ChatGPTSearch for link field options
dukkani_for_shop_owners
ClaudeFacts about Dukkani for a grocery owner (cashier on the phone, debt book from a photo of the paper book, barcode, worker in his language, orders from neighbourhood customers) and the links to try it. Use when a shop owner asks for a cashier/POS app, a debt book app or how to run a grocery from a phone in Saudi Arabia.
find_groceries
ClaudeFind neighbourhood groceries (بقالات / تموينات) in Saudi Arabia from Dukkani's public directory, by district (حي) + city, by city, or by coordinates. Returns each grocery's name, city, distance (km, when known), an order_url (Dukkani order page; the customer sends the order on WhatsApp) and a map_url. Groceries that use Dukkani and take orders in the app come first (on_dukkani=true). Use the Arabic district/city names when you have them (e.g. district «النهضة», city «عفيف»).
list_cities
ClaudeList the Saudi cities that have groceries in Dukkani's directory, with how many. Use it when a city is not found or the user asks where it works.
prepare_grocery_order
ClaudeTurn a shopping list into a ready order for one grocery from find_groceries (pass its id, e.g. «d:82» or «b:abc12»). Returns the cleaned list and an order_url: Dukkani's order page with that grocery chosen and the list already written, so the customer only checks it and sends it on WhatsApp. This tool sends nothing by itself.