Login
Check free
Login
Check free
MCP Server + Agent Skill

Let your AI agent check your AI visibility

SeenMeter ships an MCP server and an official SKILL.md so Claude Code, Cursor or Codex can pull your citations, run the audit, and write the report — without you opening a dashboard.

ClaudeCursorCodex
Join the waitlist
Run a free check first

The server

One command, four tools

No SDK, no custom client. The SeenMeter MCP server runs over stdio, like every other tool your agent already uses.

Preview — MCP ships with the app launch

$ npx seenmeter-mcp

SeenMeter MCP server v0.1.0 (preview)

4 tools registered · stdio transport

tools/list

check_visibilityRun a visibility check for a domain across all 8 engines
list_citationsList the sources AI engines cite for your tracked prompts
compare_competitorsShare of voice vs your tracked competitors, per engine
export_reportWrite the full visibility report as markdown

$

Agent tasks

What your agent can do with SeenMeter MCP

Four tools, four jobs you stop doing by hand.

check_visibility

Run the audit

Your agent kicks off a full visibility check and reads the results — without you opening a dashboard.

“Check if we're cited for 'best crm for startups' and draft fixes.”
list_citations

Pull your citations

Every source AI engines cite in your category, ready to turn into an outreach or content list.

“List the sources Perplexity cites in our category that never mention us.”
compare_competitors

Watch the gap

Share of voice against your tracked competitors, per engine, straight into your agent’s context.

“Compare our ChatGPT visibility to our top competitor this month and summarize what changed.”
export_report

Write the report

The full report as markdown — your agent can file it, post it, or rewrite it for your team.

“Run the weekly visibility report and drop it in our docs folder.”

Setup

A few lines of config

Add the server, paste your key, restart the client. That is the whole install.

Claudeclaude_desktop_config.json
{
  "mcpServers": {
    "seenmeter": {
      "command": "npx",
      "args": ["-y", "seenmeter-mcp"],
      "env": { "SEENMETER_API_KEY": "sm_live_…" }
    }
  }
}

The same server works in Cursor, VS Code, Windsurf, Codex and any other MCP-compatible client.

Agent Skill

SKILL.md — the part most MCP servers skip

MCP gives your agent access. The Skill teaches it competence.

An MCP connection alone still leaves your agent guessing: which tool for which question, what a “mention” actually means, how to structure the answer. SeenMeter ships an official SKILL.md that Claude Code, Cursor and Codex read before touching a tool.

ClaudeCursorCodex

Consistent metrics. The Skill defines mention, citation, rank and share of voice precisely — so your agent never reports one as the other.

The right endpoints. Which tool answers which question, with scope and time window spelled out. No guessing, no wasted calls.

One report format. A fixed structure for every audit, so this week’s report reads like last week’s and diffs cleanly.

ClaudeSKILL.mdPreview

## Metrics

mention ≠ citation ≠ rank — use each term only as defined below.

## Tools

check_visibility for audits, list_citations for sources, compare_competitors for share of voice.

## Scope

Always state engine, prompt set and time window before reporting a number.

## Report format

Summary → per-engine table → evidence quotes → suggested fixes.

Waitlist

Get the MCP server first

The server and Skill ship with the app launch — on every plan, including free.

Waitlist — waitlist members get access first.

FAQ

MCP + Agent Skill questions

Model Context Protocol — an open standard that lets AI agents call external tools. SeenMeter ships an MCP server, so agents like Claude Code can query your visibility data and run checks directly.

See yourself the way every AI sees you.

Free check, no signup. Then track it every day.