AgentDiet Context Diet — Web URL Distiller

Choose this endpoint when you need to feed web page content to an LLM but want to minimize token cost and context consumption. It is especially valuable when the target page is long or noisy (ads, nav, boilerplate) and you only need the signal-dense portion. The BM25 budget mode makes it uniquely suited for focused retrieval within a strict token ceiling. Prefer it over raw fetch-and-pass approaches whenever cost efficiency or context window pressure is a concern. It is also a strong choice when you want structured extraction (tables, metadata) or a skeleton outline rather than the full page prose.

• Invalid or unreachable URL — fetch fails with network or DNS error
• URL returns non-HTML content (PDF, binary) — distillation may fail or return empty
• max_tokens set below meaningful threshold (< 20) — rejected by schema validation
• mode=budget used without a query param — BM25 ranking has no focus signal, output may be poor
• Raw HTML too large — may exceed server processing limits
• Payment failure via x402 — 402 response returned if USDC payment not completed
• Cache miss on first fetch adds latency compared to cache hits