loudmink-mcp

Loudmink is an AEO (Answer Engine Optimization) platform that helps brands show up in AI search engines like ChatGPT, Gemini, Claude, Perplexity, and Grok. It tracks what those engines say about your brand, shows which sources they pull their answers from, and generates the content (articles, Reddit, YouTube) that gets your brand recommended in AI answers.

loudmink-mcp is the Model Context Protocol (MCP) server that connects Claude Code to your Loudmink account, so the model can read your AI search visibility data, manage tracked queries, and score content without leaving the terminal. It is one of several ways to work with Loudmink, alongside the web app, the loudmink-cli content tool, and the Loudmink Sync plugin for Framer.

Install

claude mcp add loudmink npx loudmink-mcp --api-key=lm_xxxxxxxx

Generate your API key in Loudmink under Settings → API Keys. Keys are scoped to one product; create separate keys per product.

You can also pass the key via env var instead of --api-key:

LOUDMINK_API_KEY=lm_xxx claude mcp add loudmink npx loudmink-mcp

To target a self-hosted instance, set --base-url=https://your-instance.example.com or LOUDMINK_BASE_URL.

What it exposes

Resource: loudmink://knowledge-base, your product knowledge base in markdown, the same context the in-app agent uses.

Read tools (10):

• list_queries: tracked queries with per-engine visibility
• get_query_detail: full visibility data for one query
• get_query_history: citation history over time
• get_digest: competitor and source analysis for a query
• get_competitors: visibility snapshot with positions
• list_articles: content library with AEO scores
• get_article: full article content and metadata
• list_reddit_opportunities: Reddit threads AI engines cite
• get_cycle_status: whether insights are currently being generated
• get_snapshots: visibility trends over time

Write tools (5):

• add_queries: add new queries to track (up to 20 at a time)
• tag_queries: add or update tags on a query
• trigger_recheck: rerun a query across all engines
• score_article: score an article for AEO compliance
• update_article: patch title or summary

Not exposed (by design): article generation, Reddit posting, content sync push, campaign triggers, agent chat. These are kept inside the web app where human review and rate limits live.

Development

npm install
npm run build       # one-off compile
npm run dev         # tsc --watch

Source: packages/loudmink-mcp/ in the Loudmink monorepo.

License

MIT