PromptbaseMCP-first
Built MCP-first for Claude, Cursor & every agent

Query millions of LLM prompts in plain English

Promptbase indexes the prompt web in real time — every prompt, every citation, every source — so your go-to-market team (or agent) can just ask.

MCP session · promptbase.query
live
"What prompts adjacent to my product are citing LinkedIn Pulse articles right now?"
Analyzing prompt graph…
Works everywhere agents live
Claude
Cursor
Windsurf
ChatGPT
Copilot
4.2M+Prompts indexed
38KNew prompts / day
12K+Sources tracked
310msAvg. query latency
Why Promptbase

Not another dashboard. A queryable prompt graph.

Everything is indexed, embedded, and cited — ready for your agent to reason over the moment it's asked a question.

Natural language queries
Ask in plain English — no SQL, no filters UI. "Show me prompts citing our competitor's pricing page this week."
MCP-native from day one
Plug straight into Claude, Cursor, or any MCP client. No dashboards to babysit — your agent does the digging.
Real-time indexing
New prompts, citations, and sources are embedded and indexed within minutes of appearing across the web.
Citation graph
Every prompt is linked to the sources it cites — Reddit threads, docs, Pulse articles — so you see influence, not just mentions.
Competitive alerts
Get notified the moment prompts adjacent to your category start citing a competitor or a new rising source.
Webhooks & API
Stream matches into Slack, your CRM, or a data warehouse the second they're indexed.
How it works

From raw prompts to a plain-English answer

01

We index the prompt web

Thousands of LLM prompts across public tools, forums, and agent transcripts, refreshed continuously.

02

You ask in plain English

Query it like you'd ask a teammate — through MCP, our API, or the web console.

03

Get ranked, cited answers

Every result comes with its source, citation graph, and relevance to your product.

MCP-first architecture

Give your agent a search engine for prompts, not a UI.

Promptbase ships as an MCP server first — the API and web console are just thin clients on top. Any agent that speaks MCP can call search_prompts, filter_by_citation, or rank_by_relevance directly.

  • Zero-config setup for Claude Desktop, Cursor & Windsurf
  • Scoped API keys per workspace and per agent
  • Streaming responses for long-running queries
mcp.config.json
{
  "mcpServers": {
    "promptbase": {
      "command": "npx",
      "args": ["-y", "@promptbase/mcp"],
      "env": {
        "PROMPTBASE_API_KEY": "pb_live_••••••••"
      }
    }
  }
}

Stop guessing what's being asked about you.

Connect your agent to Promptbase in under five minutes and start querying the prompt graph today.

Or join the waitlist