MCP Scraper is a SERP-provider integration in RankGear that stays configured under Settings but is currently held in reserve rather than selected as the active source — parked after inconsistent upstream search responses made it unreliable for present-day SERP collection.
| Term | MCP Scraper |
|---|---|
| Category | APIs, Reliability and Publishing |
| Also known as | MCP Scraper API |
| Where it appears | Settings → API keys |
What it means in RankGear
MCP Scraper sits in the provider list under Settings → API keys, where RankGear chooses which upstream service fetches live search results for its SERP work. It remains registered there, but it is not the provider currently doing the fetching. RankGear keeps the integration in place so it can be switched back on if the upstream source becomes dependable again, rather than deleting a connection that may still have a role later.
How to interpret it
Treat MCP Scraper as available-but-idle rather than broken. When you look at it, read the visible status, the diagnostics, and the credential test together before drawing a conclusion: a page that timed out points to a transient upstream hiccup, while a rejected key points to a configuration or authorization problem. The two failures call for different responses, and the on-screen signals are what separate them.
| Signal | What it tells you |
|---|---|
| Timeout or partial response | Likely transient upstream trouble; a single bounded retry may be warranted. |
| Authorization or credential rejection | Configuration problem; stop and fix the key rather than retrying paid calls. |
| Passing credential test but deselected | Integration is healthy and simply held in reserve, not active. |
Example
You open Settings → API keys to check why a batch of SERP pulls came back incomplete. MCP Scraper is listed and its read-only credential test passes, but it is not the selected provider — another source is handling collection while MCP Scraper stays parked. A timed-out request against it earns one longer, bounded retry; an authorization failure would instead halt immediately so RankGear does not burn repeated paid calls on a bad key.
Important considerations
- Retained is not the same as active: the integration being present in Settings does not mean it is feeding any current report.
- Provider search results are comparative SERP snapshots collected by a third party, not Google’s own ranking scores — they indicate what a query returned, not why a page ranks.
- Credentials and any remote publishing targets deserve care; run tests read-only where possible and keep secrets redacted.
- Distinguish a transient failure from invalid configuration before retrying, so you neither give up on a temporary blip nor repeat paid calls against a misconfigured key.
Related terms
Part of the RankGear glossary · how RankGear measures · the 870 factors.