## What we won't do

- **No CAPTCHA / bot-protection bypass.** Targets that refuse the request return
  `TARGET_BLOCKED` — this is a hard product boundary, not a missing feature.
- **Respect `robots.txt`** and site terms. LeanCrawl is for data you are allowed
  to fetch; abusive usage gets accounts suspended.

## Common questions

**Why is my crawl `partial`?** It ran out of credits. Completed pages are billed
and returned; top up and re-run with a smaller `limit` or narrower
`include_paths`.

**Why did my screenshot/PDF come back as a URL?** Artifact endpoints return a
Bearer-authenticated URL (retained for a limited window). Fetch it with the same
`Authorization` header. In MCP, artifacts are embedded inline for you.

**Do cached pages cost less?** No — a `cached` hit costs the same as a fresh
capture. The 48 h cache buys latency, not discounts.

**Where did my run go?** `GET /v1/account/activity` lists every billed attempt
with endpoint, status, `error_code`, duration, and credit cost. Content is
retained ~90 days.

**Rate limited?** You get `429` + `Retry-After`. Honor it; for heavy work use
`crawl` instead of parallel capture calls.