AnswerPeek
Log inStart free trial
CursorCodexClaudeVS Code

Your AI visibility, inside the assistant you already use

Connect AnswerPeek over MCP and let Claude, Cursor, Codex or VS Code read your scores, pull a suggested change brief, ship it — and re-check the result.

MCPhttps://www.answerpeek.com/mcp
OAuth 2.1 · dynamic registration · no API key · included on every plan
Setup · about two minutes

How to connect AnswerPeek to your assistant

1
Copy the AnswerPeek server URLYou paste this into Claude in the next step.
Server URL
https://www.answerpeek.com/mcp
2
Add it as a custom connectorIn claude.ai or the desktop app: Settings, then Connectors, then Add custom connector. Name it AnswerPeek and paste the URL.Open Claude connectors
3
Sign in, approve, askClaude sends you to AnswerPeek to sign in. Approve the permissions, then ask how your brand is doing in AI answers.Create an account

Nothing to install and no API key to store. Included on every plan, including the trial.

How it works

One question in. Real work out.

Your assistant speaks MCP. AnswerPeek exposes ten tools it can call. Ask it a question and watch it turn into work.

ClaudeAnswerPeek

Where am I losing to competitors?

Reading your visibility across the four engines
get_visibility
list_fixes

Notion is named in 14 of 24 answers (score 62). You lose “best Notion alternatives” on all four engines — Coda is named in 18. The gap: no first-party page answering the alternatives question.

What just happened
You ask in ClaudePlain language — no dashboard, no export.
get_visibility()The assistant picks the tool and calls the server over MCP.
list_fixes()Grounded in your live run data, not a template.
Answer you can act onCopy, schema and the page to change — ready to ship.
No copy-pasting numbersThe assistant reads the live data, gets the brief grounded in what the engine actually said and what your page actually contains, and writes the change in the same conversation.
The differentiator

Every other tool tells you that you’re missing. This one tells you what to write.

get_fix_brief hands your assistant the actual work: the page to change, the heading phrased the way the question is asked, the copy, the structured data. Grounded in what the engine really said and what your page really contains right now, not a template.

If your site lives in Webflow, Wix, WordPress or Shopify, connect that platform’s MCP server too and ask for the brief in cms format. Your assistant finds the problem and fixes it in the same conversation.

And then it checks itself. verify_fix answers two questions on two timescales: whether the work is live on the page, a crawl, immediate; and whether a later AI answer changed. Repeated observations can support an improvement, but they do not prove that a particular fix caused it or that it will last.

Tools

Ten tools your assistant can call

None of them can create a project, add a prompt or start a run, so the worst a connection can do is move a card.

guideRead
How the server works and the order to call things in.
list_workspacesRead
The workspaces you can work in — your own and any agency's you joined.
list_projectsRead
Every brand you track, with its score and how fresh it is.
get_visibilityRead
One brand's standing: score, usual swing, per-engine breakdown, prompts it loses, rivals named instead.
list_fixesRead
The fix board, biggest projected gain first.
get_fix_briefRead
What to actually do about one fix, for a codebase or for a CMS.
get_citationsRead
Which domains the engines cite, and what kind of place each is.
update_fix_stageWrites
Move a fix between columns, or take it off the board.
report_blockedWrites
Record why a fix could not be done.
verify_fixSpends a check
Re-crawl the page and re-measure. Uses one check.
Permissions

You approve exactly what it may do

A connection can be read-only. Verification is a separate box because it spends a check from your monthly allowance.

Read your dataanswerpeek:read
Projects, prompts, runs, fixes and citations. Every connection needs this one.
Move fixes on the boardfixes:write
Change a fix stage or take it off the board. Nothing else is writable.
Verify a fixfixes:verify
Re-crawl the page and re-measure the answer.Spends a check

Disconnect any app at any time from Settings → MCP, where these same instructions live with your connected apps beside them.

FAQ

Questions people actually ask

The Model Context Protocol is an open standard that lets an AI assistant call outside tools. Once AnswerPeek is connected, your assistant can read your visibility data and act on it instead of you copying numbers into a chat window.

Claude on the web, desktop and mobile, Claude Code, Codex, Cursor, VS Code with Copilot in agent mode, and anything else that speaks Streamable HTTP with OAuth 2.1. There is no client id to create: registration is dynamic.

No. Your client sends you to AnswerPeek to sign in and approve, the way any other OAuth app works, and receives a token bound to this server alone. Nothing to store, nothing to rotate, and you can revoke it from Settings at any time.

Only what you approved, and the list is short: move a fix between columns, take one off the board, record why one was blocked, and re-verify. It cannot create projects, add prompts or start a run. A read-only connection is a normal choice.

It spends one check from your monthly allowance, because it re-runs the prompt against the engines. That is why verification is a separate permission from writing: granting the harmless thing never forces you to grant the one that costs money.

Yes, on every plan including the trial. There is no separate charge for connecting an AI tool.

Put your visibility where the work happens

Add a brand, connect your assistant, and ship the first fix in the same conversation.

Start free trial Try a free check
Free for 7 days · no card · MCP included on every plan