Somewhere in your Search Console account is a list of pages that rank eleventh. They already have impressions, Google already thinks they are relevant, and almost nobody clicks them, because almost nobody reaches the second page. Moving one of those up two places is the cheapest traffic in SEO, and the reason it goes unclaimed is that the list is tedious to build in the Search Console interface.

It is a single query if you can ask for it directly. Model Context Protocol lets your AI editor call Search Console for you, so "show me the pages sitting just off page one" is one question instead of five clicks, a date range and a copy paste.

What striking distance means

A striking distance page is one ranking roughly between position 8 and 20 for a query that already brings it impressions. It is close enough that a title rewrite, a better meta description or a section that answers the query directly can move it onto page one, and far enough down that it currently earns very few clicks. The list is derived, not stored: you build it by querying impressions and average position together, then sorting by impressions so the biggest opportunities surface first.

Two numbers decide whether a row is worth your afternoon. Impressions tell you how much traffic exists to win. Average position tells you how far you have to move. A page at position 11 with 900 impressions is worth more attention than a page at position 4 with 40, and the Search Console interface will not sort by that combination for you.

Click through rate is the third column worth reading, because it separates two different problems that look alike. A page with impressions and a near zero click through rate at position 6 does not have a ranking problem, it has a title problem. A page at position 15 has a ranking problem. The fix is different in each case, and the numbers tell you which one you are looking at.

The four tools, and which question each answers

  • gsc_analytics. "What is happening with my traffic?" Queries the Search Analytics API with a date range, dimensions and filters. Returns impressions, clicks, click through rate and average position. This is the tool that builds the striking distance list.
  • gsc_site_list. "Which properties do I have?" Enumerates the properties your Google account can reach. Usually the first call in a session, to scope the rest.
  • gsc_inspect. "Is this URL indexed?" Wraps the URL Inspection API and returns indexing status, last crawl date and page experience signals. The check to run after you republish an improved page.
  • gsc_sitemap_list. "What sitemaps have I submitted, and are they healthy?" Lists submitted sitemaps with submission date, last processed date and error count. The fast check before assuming an indexing problem is on Google's side.

What Search Console will not tell you

Search Console reports your own property. It has no search volume for a keyword you do not rank for, no keyword difficulty, no backlink index, no competitor data and no site crawl. A vendor backed suite such as OpenSEO covers those and ToolPiper does not; OpenSEO is open source under MIT, and its data is metered through DataForSEO. Verified 2026-08-28. What Search Console does have is the one dataset none of them can sell you, which is what Google actually showed for your pages.

How do I set up Google Search Console with ToolPiper?

Connect Google in ToolPiper's OAuth pane. Register ToolPiper with your AI editor. Ask the editor a Search Console question in plain English. The editor calls the right tool with the right scope based on the prompt.

Three steps from zero:

  1. Connect Google. Open ToolPiper. OAuth pane. Click Connect on Google. Sign in with the account that owns your GSC properties. Grant the webmasters.readonly scope ToolPiper requests. The OAuth pane updates to show "Connected". This is the only setup specific to Search Console. The rest is generic MCP.
  2. Register ToolPiper with your AI editor. Follow the matching guide for your editor: Claude Code, Cursor, Claude Desktop, Devin. Any MCP-capable editor works.
  3. Ask a Search Console question. Start a chat. Try "which pages on my site dropped impressions in the last 7 days?". The editor calls gsc_site_list, picks the property, calls gsc_analytics with the right date range and dimensions, and summarizes the result.

There's nothing to authenticate by hand on the editor side. Your editor reaches ToolPiper over loopback, where ToolPiper mints a short-lived bearer token (rotated every launch, stored at ~/Library/Application Support/ToolPiper/.toolpiper-token with 0600 permissions) and trusts the connection at the socket level. Only off-machine clients (a LAN device or a PiperMesh peer) ever need a scoped Bearer tp_<64 hex> token. Once your editor is connected it shows up in ToolPiper's Connected Apps pane, where you can see its tool access and revoke it in one click. If you want to keep the GSC tools but block something heavier like system_run_command for every client, the tool-governance overlay does that: a per-device, locally-edited allow/block layer today that an organization can push centrally later.

What kinds of questions can I ask?

Anything Search Console can answer about your site's organic traffic, plus comparison and inference questions an AI is better at than the GSC UI. The editor handles the API parameter shape. You ask in plain English.

Example prompts that work end-to-end with the four tools:

  • "Compare impressions and clicks between this week and last week. Which pages changed the most?"
  • "Is the article at /blog/connect-toolpiper indexed yet? When was Google's last crawl?"
  • "What queries did I rank in the top 3 for last month? Group by query intent."
  • "List my top 10 declining pages by impressions, week over week."
  • "Did my last sitemap submission process successfully? Any errors?"
  • "Which of my pages have impressions but no clicks? Suggest a title or meta description fix for each."

The last example is where AI editors pull ahead of the GSC UI. They can fetch the analytics, fetch the page content separately, and synthesize a recommendation. The GSC UI can't.

How does ToolPiper handle Search Console API rate limits?

Search Console has generous quotas for individual property owners (1,200 queries per minute per user). ToolPiper passes through your account's quota directly without batching or rate-limiting on its end. For most users this is invisible. For high-volume scripted queries, ToolPiper surfaces Google's quota errors when they happen.

If you're running a lot of gsc_analytics queries in a tight loop, you may hit Google's per-minute limit. The error comes back as an MCP tool error with Google's message. Wait and retry, or break the query into fewer larger ones (longer date ranges, more dimensions per call).

What if Search Console tools return permission errors?

Three causes account for most permission errors. The connected Google account doesn't own the property. Workspace admin policies restrict third-party app access. The OAuth token expired and ToolPiper hasn't re-authenticated. Check the OAuth pane status first.

  1. Property ownership. The Google account you connected has to own (or be granted access to) the GSC property you're querying. gsc_site_list only shows properties this account can see.
  2. Workspace policy. On Google Workspace accounts, admins can restrict third-party app access. If ToolPiper isn't approved by your admin, the consent flow may succeed but tool calls fail with permission errors. Try a personal Google account first to isolate the issue.
  3. Token expiry. If the OAuth pane shows "needs re-authentication", click Connect again. ToolPiper refreshes tokens automatically, but some policy changes invalidate refresh tokens and force a re-consent.

For deeper OAuth troubleshooting, see Connect ToolPiper to Your Cloud Accounts via OAuth on Mac.