Rankealo
Start

Rankealo for your AI agent

Give Claude Code a set of ready-made Rankealo workflows. Each skill turns one job — diagnose AI visibility, close a content gap, triage a GEO audit — into a cost-conscious sequence of MCP tools, so you get a diagnosis and next actions, not a raw data dump.

What you get#

The Rankealo MCP server gives your agent native tools over your domain. The agent skills layered on top teach the agent how to use those tools well — which to call, in what order, when to stop, and what each one costs. They sell visibility outcomes, not data access, and are not a rank tracker.

1. Install the MCP server#

From your project directory (or use the in-app setup at /mcp):

bash
npx @rankealo/mcp-server mcp init

The init flow writes MCP config for Claude Code, Cursor, Windsurf, or Codex and prompts for your domain-scoped API key. The server expects:

env
RANKEALO_API_KEY=rnk_your_key_here
RANKEALO_BASE_URL=https://app.rankealo.ai

Read vs write keys

A read-scoped key runs every skill on this page. Write actions (publish, reschedule, status, settings) need a write-scoped (Growth) key — read keys receive 403, and the skills tell you when that happens.

2. Install the skills in Claude Code#

The skills ship inside the @rankealo/mcp-server package under skills/. Copy the folders into a directory Claude Code loads — project-scoped (.claude/skills/) or user-scoped (~/.claude/skills/):

  1. 1

    macOS / Linux

    bash
    # resolve the installed package path, then copy the skill folders
PKG=$(npm root)/@rankealo/mcp-server
mkdir -p ~/.claude/skills
cp -r "$PKG"/skills/rankealo-* ~/.claude/skills/
  2. 2

    Windows (PowerShell)

    powershell
    $pkg = Join-Path (npm root) '@rankealo/mcp-server'
New-Item -ItemType Directory -Force $HOME\.claude\skills | Out-Null
Copy-Item -Recurse "$pkg\skills\rankealo-*" $HOME\.claude\skills\
  3. 3

    Confirm

    Start Claude Code in the project. Ask something like “run a Rankealo visibility checkup” — the matching skill loads by its name and drives the MCP tools for you.

The six skills#

SkillJob it does
rankealo-visibility-checkupWhere is my brand losing ground across AI engines, and what do I fix first?
rankealo-content-gapTurn competitor gaps into proposed article topics.
rankealo-keyword-pool-reviewAudit keyword-pool health and suggest research directions.
rankealo-geo-audit-triageExplain each failing GEO/AI-readiness check and order the fixes by impact.
rankealo-article-briefAssemble a data-grounded brief for one keyword.
rankealo-weekly-reviewThe Monday-morning digest: visibility delta, content shipped, regressions, next actions.

All six are read-only and run as a single bounded pass. For raw copy-paste MCP call sequences, see Agent recipes.

What each call costs#

  • Free. Every get_*, list_*, and compare_* tool reads already-synced data. It never triggers a live measurement; freshness reflects the last sync (check with get_data_freshness).
  • Spends credits. chat_seo_agent runs the Rankealo SEO agent, an LLM call metered against your plan. None of the six skills call it by default.
  • Writes. publish_article, reschedule_article, set_article_status, and the update_* settings tools change state and need a write-scoped key. The skills read current state and confirm with you first.

Large jobs

The server asks your agent to confirm before running large batches (many per-page breakdowns, fanning a tool over dozens of URLs). Prefer one bounded pass over polling loops.

Still stuck?