Webpage Distiller with BM25 Budget Control

Use this endpoint when an AI agent needs to consume web content that would otherwise overflow its context window, or when you need BM25-ranked relevance filtering to extract only the portions of a page relevant to a specific query. It is especially valuable over raw fetch-and-pass approaches when token cost and context efficiency matter. Prefer the budget mode for precise token-ceiling enforcement, structured mode when you need tabular or metadata extraction, and skeleton mode for fast triage of long documents.

• URL is unreachable or returns non-200 status — fetch failure error
• HTML payload is empty or malformed — parse error
• max_tokens set too low (below 20) — validation error rejected by schema
• mode=budget used without a query field — may return unguided excerpt
• Rate limit or payment failure via x402 — 402 Payment Required response
• Target site blocks crawlers — may return empty or partial content