Creativity · MCP — server

MCP Honeycomb Server

Honeycomb is the observability platform for high-cardinality debugging. The MCP Honeycomb Server wraps its query and BubbleUp APIs so an LLM client can run queries across wide events, inspect SLOs, and explore recent triggers — effectively letting Claude BubbleUp alongside an on-call engineer.

MCP facts

Kind: server
Ecosystem: anthropic-mcp
Language: TypeScript / Node.js
Transports: stdio

Capabilities

Tools: list_datasets, run_query, list_slos, list_triggers
Optional BubbleUp helper that returns ranked dimension differences
Region-aware (US / EU)

Install

npx -y @modelcontextprotocol-community/server-honeycomb

Configuration

{
  "mcpServers": {
    "honeycomb": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol-community/server-honeycomb"],
      "env": {
        "HONEYCOMB_API_KEY": "<config-api-key>",
        "HONEYCOMB_REGION": "us"
      }
    }
  }
}

Frequently asked questions

Can the model create dashboards?

Honeycomb's API supports board creation; most MCP builds keep it read-only. Creating a board from a chat request is possible but gated by policy in most teams.

How does BubbleUp integrate?

Give the model a slow-request query; the BubbleUp tool returns the dimensions that differ most between slow and fast events, which Claude can summarize in plain English.

Event volume concerns?

Honeycomb dynamically samples data, so the MCP server doesn't need to page through millions of events — it reads the aggregated result.

Sources

Honeycomb API Documentation — accessed 2026-04-20
Model Context Protocol — accessed 2026-04-20