search_content
ChatGPTALWAYS use this tool to search the knowledge base when the user asks ANY question. This tool searches for relevant content using semantic similarity. Use it for questions about products, policies, pricing, documentation, features, how things work, troubleshooting, and any other information. For complex questions covering multiple aspects, use multi-query: query: [{ query: "pricing plans", aspect: "pricing" }, { query: "main features", aspect: "features" }]. Recency control (dateRange + sortByDate) — BY DEFAULT, OMIT BOTH: - Most queries are evergreen / definitional / how-to and don't need recency. "Mi a Bitcoin?", "What is SWIFT?", "How do I configure X?" → call WITHOUT dateRange or sortByDate. - ONLY fill these inputs if the user's query contains an explicit recency cue: Hungarian triggers: "legfrissebb", "friss", "legutóbbi", "mostani", "idén", "tavaly(i)", "ebben az évben", "elmúlt N nap/hónap", "<YYYY>-ben" English triggers: "latest", "most recent", "newest", "this year", "last N days/months", "in <YYYY>", "since <date>" - Mapping (compute date math from the current date in your system prompt): "Mi a legfrissebb cikk?" / "What's the latest article?" → sortByDate: "desc", limit: 1 "Az elmúlt 30 nap hírei" / "News in the last 30 days" → dateRange: { after: "<today - 30d>" } "2025-ben publikált anyagok" / "Articles published in 2025" → dateRange: { after: "2025-01-01", before: "2025-12-31" } "Idén / this year" → dateRange: { after: "<Jan 1 of current year>" } * "Régebbi archív cikkek" → dateRange: { before: "<today - 2y>" }, sortByDate: "desc" Never answer without searching first.
