Summary
The page that would not fetch, without standing up a browser stack for it.
The escalation is explicit rather than automatic: basic mode is the fast path and the stealth modes are what you reach for once it fails, so you are not paying browser costs on every request. On a long page s_fetch_pattern is the difference between the matches with their surrounding context and the megabyte they were sitting in.
What it is
A fetcher with three escalating modes for sites that run bot detection. One tool returns the page with pagination, the other returns only the regions matching a regex, each with context around it.
What you get
- A full page fetched with pagination through start_index and max_length, so a long document arrives in pieces rather than cut off
- Regex-matched extraction with a context_chars window around each match
- Three modes to escalate through — basic first, then stealth or max-stealth when basic fails
- A METADATA header on every response carrying length and truncation status, or match statistics on a pattern fetch
Requirements
Uv on your PATH. No account and no key.
Setup effort
One command — uvx scrapling-fetch-mcp