AskWatch Google Search Console MCP
officialAsk Google Search Console in plain language. Hosted, free, no Google Cloud project.
What can you do with AskWatch Google Search Console MCP?
-
Site performance overview — Ask how your site did last month and get clicks, impressions, CTR, and position versus the prior period via
site_overview. -
Spot traffic drops — Ask what dropped in the last 28 days and which pages or queries lost the most, using
changes. -
Find ranking opportunities — Ask for queries ranking 5–20 with below-average click-through rates to target for improvement, via
find_opportunities. -
Detect keyword cannibalization — Ask which queries pull up two of your pages at once and how each performs, using
find_cannibalization. -
Check page visibility — Ask what searches a specific page (like /pricing) is shown for, or whether a URL is indexed and when it was last crawled, via
page_queriesorinspect_url. -
Analyze branded vs. non-branded traffic — Ask how much of your traffic is branded versus non-branded, split on your own spellings, with
brand_split.
Hosted MCP Server
npx add-mcp 'https://mcp.askwatch.ai/gsc'Installs into Claude Code, Codex, Cursor and more
Documentation
Free Google Search Console MCP
Ask Search Console in plain language from Claude, ChatGPT, Cursor or Claude Code. Connect your Google account once and we hand you a link on mcp.askwatch.ai. No Cloud project, no credentials.json, nothing to run.
- Read-only access.
- Free, no card.
- Revoke in one click.
Claude
askwatch-gsc
what dropped in the last 28 days?
askwatch-gsc · changes
Clicks are down 11% against the previous 28 days. Three queries account for most of it, and all three lost position rather than impressions.
Clicks
1,284
Impressions
48,910
Avg position
14.2
- invoice template word9.1 → 14.6
- how to invoice a client7.4 → 11.2
- free invoice generator12.0 → 15.3
Period 25 Aug to 21 Sep, data through 19 Sep. Position is averaged across impressions.
Connected in about a minute
Three steps, and none of them involve a Google Cloud console.
- Step 1
- Step 2
Connect Google Search Console
One consent screen, read-only. We never see your Google password, and the grant is yours to withdraw. - Step 3
Paste your link into your AI client
Your own address, and the only thing your client needs. Anyone holding it can read the same data, so keep it as you would a password.
Chat
Editors
Terminal
- 1.Open Settings and go to Connectors.
- 2.Choose Add custom connector.
- 3.Paste your link as the server URL and save.
https://mcp.askwatch.ai/gsc/your-personal-keyCopy
Then ask: How did my site do last month?
No Google Cloud project, no credentials.json
Every Search Console MCP you run yourself starts in a developer console. This one starts with a consent screen.
Running one yourself
What the setup asks of you before the first question.
- Create a Google Cloud project
- Enable the Search Console API on it
- Configure an OAuth consent screen and add yourself as a test user
- Download credentials.json and keep it on your disk
- Run the server yourself, and keep it running
AskWatch, hosted
Free, and the credentials never touch your machine.
- Sign in to AskWatch with your email
- Approve one Google consent screen
- Paste the link, and the server is ours to run
Things worth asking Search Console
Type the question. The client picks the tool, the dates and the thresholds, and answers from your own data.
- “How did my site do last month?”
site_overview - “What dropped in the last 28 days, and on which pages?”
changes - “Where am I ranking 5 to 20 with a click-through rate below what that position usually gets?”
find_opportunities - “Which queries are pulling in two of my pages at once?”
find_cannibalization - “What searches is /pricing actually shown for?”
page_queries - “How much of my traffic is branded?”
brand_split - “Is this URL indexed, and when was it last crawled?”
inspect_url
Eleven Search Console tools
Seven that answer a whole question with the analysis already in them, one that passes anything through, three that say what you have.
| Tool | What it answers | Kind |
|---|---|---|
site_overview | Clicks, impressions, CTR and average position for a period, each against the period before it. | Question |
top | Top queries, pages, countries or devices, optionally narrowed to one country or device. | Question |
changes | What gained and what lost between two periods, with rows too small to mean anything filtered out. | Question |
find_opportunities | Queries ranking 5–20 whose click-through rate is below what that position usually earns. | Question |
find_cannibalization | Queries where more than one of your pages is being shown, and how each one does. | Question |
page_queries | What a specific page is actually shown for. | Question |
brand_split | Branded against non-branded search, split on the spellings you supply. | Question |
query | The Search Analytics query itself, with the dates, dimensions and filters Google takes, for questions the seven scenarios do not shape. | Raw |
list_sites | The properties this link can read, and your level of access to each. | Account |
inspect_url | Whether a URL is indexed, and what Google saw when it last crawled it. | Account |
list_sitemaps | The sitemaps a property has submitted, with their errors and last read. | Account |
Numbers you can forward
The reader here is a model, and a model repeats what it is given. So every answer carries the things a careful analyst would have said out loud.
-
Every answer names its window
The period it covers and the last day Google has finished counting, in the answer itself, rather than left for the model to assume. -
The unfinished day is left out
Search Console closes a day late, and a partial day dragged into a comparison once reported an 11.7% fall in clicks on a site that had fallen 1.1%. We ask Google for finalised data only. -
The method travels with the numbers
Which window, which threshold, which sort. A model that can read the criteria can repeat the question and explain the answer. -
Google's own caveats come along
That position is averaged across impressions, and that some queries are withheld for privacy, ride along with the rows instead of sitting in a help page nobody opens.
What it can reach, and what we keep
A link that reads your Search Console is worth being careful with. This is what it can do, and how to end it.
-
Read-only, at Google's level
The grant asks for webmasters.readonly. It cannot submit a sitemap, request indexing, remove a URL or change a property, and not by our choice at request time: the scope itself does not allow it. -
Tokens encrypted, link stored as a hash
Your Google tokens are encrypted before they are stored. The link is kept as a hash, so what sits in our database is not a working address to anybody who reads it. -
Revoke from either side
Regenerate the link and the old address stops answering immediately, telling whoever holds it that it was replaced. Or withdraw the grant from your Google account and the whole connection ends, ours included. -
What we keep, and for how long
The properties your account can read, the answers Google gave (cached from thirty minutes to thirty days, by how old the data is, so one question does not spend your quota twice) and a count of calls per tool per day. Disconnecting drops the grant and the cache read under it. -
Limits guard, they do not ration
20 calls a minute is the only refusal, and it is there to stop a looping agent from spending your Google quota. The daily thresholds add a notice and a small delay; nothing is ever cut off. -
Not sold, not trained on
AskWatch's use of information received from Google APIs follows the Google API Services User Data Policy, including the Limited Use requirements. We do not sell this data and we do not train models on it. What we store and for how long is in our Privacy Policy; the terms of the connection are in clause 1.9.
Frequently asked questions
What is the AskWatch Google Search Console MCP server?
It is a free hosted MCP server that lets an AI client read your Google Search Console data. You connect your Google account once, AskWatch issues you a personal link, and you paste that link into Claude, ChatGPT, Cursor or Claude Code. From then on you ask questions in plain language (how the site did last month, what dropped, which pages compete with each other) and the client answers from your own Search Console data. Access is read-only.
Do I need a Google Cloud project or a credentials.json file?
Which AI clients does it work with?
Is it really free?
What can it change in my Search Console?
What do you store?
How do I revoke access?
Does it work with domain properties?
Why is the last day or two missing?
Are there rate limits?
Other free tools
-
AI Visibility Checker
Check if ChatGPT, Gemini, Perplexity and Google AI name your brand, and get your AI visibility score. Open tool -
AI Overview Checker
See if Google AI Overviews cite your site, which of your pages they pick, and who is cited instead. Open tool -
Perplexity Visibility Tracker
Track whether Perplexity cites your site and names your brand in its answers. Open tool -
ChatGPT Visibility Tracker
See if ChatGPT mentions your brand and where you rank in its answers. Open tool -
Gemini Visibility Tracker
Check if Google Gemini mentions your brand and who it names instead. Open tool -
Google AI Mode Visibility Tracker
See if Google AI Mode names your brand and which pages it cites. Open tool -
Copilot Visibility Tracker
See if Microsoft Copilot mentions your brand and cites your site. Open tool
Put Search Console where you already ask questions
One consent screen, one link, and your next question about the site gets a real answer.
- Read-only access.
- Free, no card.
- Revoke in one click.