AI Visibility MCP for Cline

In Cline, your agent can pull the raw answers the six engines actually give - ChatGPT, Perplexity, Google AI Overviews, Google AI Mode, Copilot and Gemini. Ask in plain language and the answer text and citations come back as a structured tool result, in the same VS Code window as the docs and landing-page files that need fixing. One MCP server, one narrow tool.

Connect AgentGEO to Cline

Cline configures MCP servers in a JSON settings file. Open the MCP Servers panel and add an agentgeo entry to cline_mcp_settings.json:

{
  "mcpServers": {
    "agentgeo": {
      "command": "npx",
      "args": ["-y", "agentgeo-mcp", "--key", "ag_live_..."]
    }
  }
}

A live ag_live_ key comes with a plan. Want to try the wiring before spending anything? An ag_test_ key always returns clearly labelled demo records at zero credits. Reload, and the fetch_raw_answers tool appears in Cline's tool list.

Every call returns one record per surface - the provider's answer, verbatim, with every source it cited as structured data:

{
  "surfaceKey": "perplexity",
  "status": "delivered",
  "answerText": "...",
  "sources": [
    { "title": "...", "url": "https://...", "position": 1 }
  ],
  "fetchedAt": "2026-07-14T09:12:04Z",
  "latencyMs": 8412,
  "providerRecordId": "..."
}

What to ask once it's connected

• "Fetch what chatgpt and perplexity answer for our main category query and tell me whether any cited domain is ours." The baseline visibility check, without opening a browser tab.
• "Pull all six surfaces for the query our homepage targets, diff the sources[] lists, and write the result to geo-notes.md." The evidence lands in the repo it's about.
• Use plan mode to have Cline compare Perplexity's citations against the pages under docs/ and propose which ones to tighten - then switch to act mode and let it edit the markdown.
• "Check whether our pricing page appears in google_ai_overview sources for our product-plus-pricing query; if not, rewrite the page's opening to answer it directly."
• "Diff gemini against copilot for the same query and list the competitor domains cited by both."
• After the fixes ship, re-fetch the same query and diff answerText and sources[] against last week's record - check, fix, verify, all in one window.

What the MCP server gives you

• One narrow tool - fetch_raw_answers. No tool sprawl in your agent's context; one contract to learn.
• Six engines behind one contract - the same arguments and record shape for every surface, so widening a check is one more string in surfaces.
• Structured sources[] - title, URL and position for every citation, ready to diff, store or grep. No regexing links out of prose.
• Raw records only - no rankings, sentiment or visibility scores baked in. The conclusions stay in your agent, next to your context.
• Managed-scraper engine, maintained for you - the server talks to AgentGEO's API; you never babysit a headless browser or rotate proxies.
• Usage billing with a spend cap - one credit per delivered record, never per-seat.

What it doesn't do

AgentGEO is deliberately a GEO data layer, not a platform. The analysis belongs in the agent you already trust with your repo.

• No dashboards, no scores, no recommendations engine. Cline - with whatever model you brought - does the ranking, diffing and fixing. That's the point.
• Not instant. Live surfaces are slow: requests wait up to 180 seconds, and an AI Overview SERP round-trip alone runs 40-90s. Let the tool call finish.
• Some fetches finish async. A slow chatbot scrape can exceed the sync budget and fail with a providerFields.snapshot_id; retry with that id and the same single surface to collect the finished answer without paying for a re-scrape (not valid for google_ai_overview).

Prefer the REST API?

The same records are one HTTP call away: POST https://api.agentgeo.org/v1/fetches. See the Python and curl pages for the request shape - useful when a check graduates from an in-editor ask to a script committed next to your site.

Who builds on this

Developers who run Cline because it's open source and lets them bring their own model - and who edit their site or docs repo in the same VS Code window. If your GEO todo list is a stack of markdown files and your instinct is check it, then fix it, then commit it, wiring the data layer into the editor beats logging into someone else's dashboard.