Kiwi

Kiwi for Honeycomb

Have the agent on a Kiwi step query Honeycomb traces and run BubbleUp on slow requests, through Honeycomb's MCP server.

  • Your agent works in Honeycomb through its own connection. Kiwi does not call Honeycomb and never holds its credentials.
  • A workflow skill can list Honeycomb among the tools it requires. When you supply your agent's available tools to a skill check and Honeycomb is not among them, the check reports SK15-UNSUPPORTED-TOOL.
  • A project Admin can add links to Honeycomb, with notes, under Connections and bind them to the steps that use them. The agent reads them with list_connections.
  • The agent can report the tools and plugins it ran with, including its Honeycomb connection, with report_configuration.
  • The agent submits links to its work in Honeycomb as the step's outputs. People review them, and sign-off stays with the people your workflow names.

What your agent can do in Honeycomb

  • Run a query against a dataset and find the columns to group or filter by.
  • Run BubbleUp to see which attributes set outlier events apart from the rest.
  • Fetch a trace and read the details of its spans.
  • Read the service map, boards, triggers and SLOs of an environment.

Set up

  1. Connect your agent to Kiwi

    Add Kiwi's hosted MCP server, https://kiwiflow.io/mcp, to your coding agent.

  2. Connect your agent to Honeycomb

    Add the Honeycomb MCP server to your agent with this URL. The credentials stay with your agent.

    https://mcp.honeycomb.io/mcp
  3. Example skill

    Publish instructions like these as a workflow skill that lists Honeycomb among its required tools.

    Start from the endpoint and the latency target given in the run's inputs.
    In Honeycomb, query that endpoint's spans for the last 24 hours and run
    BubbleUp on the requests slower than the target.
    Submit the queries you ran and the attributes BubbleUp found as outputs.
    If the dataset has no spans for the endpoint, report a blocker to the Owner.