Connect the Honeycomb MCP server to Unblocked to give AI agents live tracing and observability context

High-cardinality questions — which customers, which endpoints, which build — are where Honeycomb shines, but answering them means opening the query builder, picking a dataset, and digging through a trace waterfall by hand. A coding agent can’t do any of that on its own. Honeycomb’s MCP server lets AI agents query and explore telemetry the way you do in the Honeycomb UI, and Unblocked connects to it once so every agent can ask without its own Honeycomb setup.

This is a third-party MCP server: Honeycomb hosts and maintains it, and Unblocked connects to it as a verified MCP server. Nothing is ingested or indexed — Unblocked runs the read-only queries, trace lookups, and BubbleUp comparisons you allow when a question needs them, so results come from your events as they stand, not a snapshot.

Wiring Honeycomb’s MCP server straight into an agent gives it query results and span lists. Unblocked’s Context Engine reasons across those results and your other sources — following a BubbleUp outlier on a slow span to the code path that emits it, the pull request that changed it, and the ticket where the slowdown was reported — and returns one cited answer.

Once connected, any MCP-compatible coding agent — Claude Code, Cursor, Devin, GitHub Copilot, Codex — reaches Honeycomb context through a single Unblocked MCP server connection.

What you get

• Live queries, traces, and BubbleUp — Agents run time-series queries, fetch full trace waterfalls, and compare outliers against a baseline with BubbleUp — all against your current Honeycomb data.
• SLO and Trigger awareness — Agents read SLO status and Trigger configuration, so a question about reliability starts from what your team already measures and alerts on.
• Read-only, tool-level control — Tools that create or update Boards, Triggers, SLOs, or recipients are never offered. You choose which read-only Honeycomb tools Unblocked may call from Settings → Data Sources.
• Context Engine reasoning — Honeycomb results are reconciled with your code, pull requests, tickets, and conversations, so an anomalous span leads to the change behind it — with citations.

Prerequisites

• An Unblocked account (free to start).
• A Honeycomb team with Honeycomb Intelligence enabled, and access to the environments you want agents to query so you can authorize the OAuth connection.
• An MCP-compatible coding agent (Claude Code, Cursor, Devin, etc.) to use the context in your editor.

Step 1: Connect Honeycomb to Unblocked

1. In Unblocked, go to Settings → Data Sources → Connect another data source, and under Incident Management select Honeycomb.
2. The Honeycomb MCP server URL is filled in for you. Choose OAuth, click Connect, and sign in to Honeycomb to authorize Unblocked.
3. Once Unblocked discovers the server’s tools, select the read-only tools it may use and click Save Settings.

Step 2: Install the Unblocked MCP server

On macOS or Linux, one command auto-detects and configures the supported agents on your machine:

On Windows, or when you prefer remote MCP, add the server manually using the configuration below, or follow the manual install guide .

claude mcp add -s user --transport http unblocked https://getunblocked.com/api/mcpsse

{
  "mcpServers": {
    "unblocked": {
      "url": "https://getunblocked.com/api/mcpsse"
    }
  }
}

https://getunblocked.com/api/mcpsse

Honeycomb hosts and maintains this MCP server. Unblocked calls only the read-only tools you select, at the moment a question needs them:

• run_queryRuns a time-series aggregation query against a dataset and returns the results.
• get_traceRetrieves every span for a trace ID and returns it as a waterfall.
• run_bubbleupCompares a selected subset of data against the baseline and returns what sets it apart.
• find_columnsFinds columns and calculated fields that match a natural-language intent, so queries use real field names.
• get_slosLists SLOs, or returns one SLO’s status and graphs.
• get_triggersLists Triggers, or returns one Trigger’s configuration.

To give Claude Code, Cursor, GitHub Copilot, Codex, or another MCP-compatible AI coding agent access to Honeycomb context, connect Honeycomb to Unblocked and add the single Unblocked MCP server to your agent.

Unblocked acts as the MCP context layer for your Honeycomb data, searching traces, queries, and SLOs alongside your other connected sources and returning cited results through these tools:

• context_researchResearches a question across your connected sources and calls the Honeycomb tools you selected when traces, queries, SLOs, or Triggers are relevant, returning a synthesized answer with citations.
• context_get_urlsFetches the full content behind URLs your agent encounters across connected sources — Jira, Linear, Notion, Confluence, Slack, GitHub, and more.

Questions you (or your agent) can ask Unblocked about your Honeycomb data:

• Which endpoints got slower after yesterday’s release, and which pull request touched them?
• Run BubbleUp on the slow checkout spans — what do they have in common, and is there a ticket about it?
• Is the API availability SLO burning faster this week, and what did the team say about it in Slack?
• Show me the trace for this request ID and point me to the code that emits its slowest span.