# Honeycomb MCP (slim)

> Honeycomb's hosted MCP server lets AI agents query a team's traces, metrics and events, run BubbleUp analysis, and create or update Boards, Triggers and SLOs. It uses OAuth or a Management API key and has US and EU endpoints.

- Full: https://www.anchorterminal.com/tools/honeycomb-mcp.md (~8,200 tokens) · this version ~1,830 tokens · JSON https://www.anchorterminal.com/tools/honeycomb-mcp.json · canonical https://www.anchorterminal.com/tools/honeycomb-mcp
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · Updated: 2026-10-09

**C · 58.3/100 · rank #527 of 842 · #3 in Observability & incidents · not agent-ready · confidence medium**


Assessment: OAuth with `mcp:read` and `mcp:write` scopes keeps the 34 tools read-only until write access is granted, and per-tool rate limits are published. Tool input schemas are visible only after sign-in, and no injection guidance was found for telemetry content. The product changelogs disallow unnamed crawlers, so release history went unread.

## Facts

- Kind: MCP server · vendor: Honeycomb · category: Observability & incidents · legal entity: Hound Technology, Inc. d/b/a Honeycomb · provenance 71/100
- Endpoint: `https://mcp.honeycomb.io/mcp` (Streamable HTTP)
- Auth: OAuth or key · pricing: Your plan · x402: no · licence: Proprietary hosted service under Honeycomb's Terms of Service and Honeycomb Intelligence Supplemental Terms. The companion plugin in honeycombio/agent-skill is MIT
- Probe metrics: not measured yet (probes haven't run)
- What is graded: The hosted remote MCP server at https://mcp.honeycomb.io/mcp (EU https://mcp.eu1.honeycomb.io/mcp). It is closed source. The MIT plugin in honeycombio/agent-skill adds skills, two agents and a setup command, and is not required
- Tools: 34 in the public reference. Discovery (4), query and analysis (5), traces and spans (3), service map and anomalies (2), semantic conventions (3), Boards (3), Triggers (3), SLOs (3), recipients (2), Canvas (2), AI conversation analysis (2), `refinery_docs` and `feedback`. `get_service_map` is Enterprise only
- Prompts: MCP prompts such as `honeycomb-onboarding`, `honeycomb-learning` and `otel_analysis`, shown by most clients as slash commands. They need no extra scope
- Credentials: OAuth authorisation code with PKCE S256 and refresh tokens, scopes `mcp:read` and `mcp:write`, token in the header only. Or a Management API key sent as `Bearer <KEY_ID>:<SECRET_KEY>`, which only a Team Owner can create and whose scopes are fixed at creation
- Tool subsets: `X-MCP-Toolset` header with `all` (default), `query` or `canvas`. The header narrows what loads and grants nothing
- Rate limits: Per tool, per minute. 300 for semantic-convention tools and `canvas_agent_poll_response`, 200 for discovery tools, 150 for `get_triggers`, `find_queries` and `get_query_results`, 100 for `run_query`, `get_trace`, `get_dataset_columns` and `get_slos`, 30 for updates, 20 for creates, 10 for `get_service_map`, 50 otherwise. The docs say limits may change
- Errors: Rate-limit errors read `Rate limit exceeded, please try again after <time>`. The troubleshooting page also covers Unauthorized, missing tools, missing `mcp:write` and unknown environments, datasets or columns
- Sessions: MCP sessions may time out after 24 hours. OAuth sessions are listed on the account's Integrations tab
- Plans: Honeycomb MCP is in the Free, Pro and Enterprise feature lists. Free allows 20 million events and 100 million metric data points a month. Pro starts at $150 a month for 50 million events. Enterprise is quoted
- Requirement: A Team Owner must enable Honeycomb Intelligence for the team. Through MCP the Activity Log environment is shown only to Team Owners
- Status: Generally available. The docs list MCP among past betas as graduated. Registry entry `io.honeycomb/mcp` 1.0.0, published 29 April 2026
- Last dated change read: The tools reference was last modified on 29 September 2026 per the docs sitemap, and the plugin reached 1.2.2 on 20 September 2026. The product changelogs were not read
- Prices: Pro plan, 50 million events $150 per month (plan)

- Scores: Reliability 65, Performance pending, Schema & documentation 57, Agent ergonomics 53, Security & auth 71, Payments & pricing 30, Task success pending, Maintenance & community 61, Transparency & trust 68 · total over the 7 assessed categories
- Why: Reliability, Scored as a hosted MCP server. · Schema & documentation, The server is closed source and `tools/list` needs sign-in, so we could not read the input schemas. · Agent ergonomics, 34 tools is above the 30 line (5). · Security & auth, OAuth authorisation code with PKCE S256, dynamic client registration, client ID metadata documents, refresh tokens and the scopes `mcp:read`… · Payments & pricing, No x402, MPP or L402 (0). · Maintenance & community, The product changelogs at changelog.honeycomb.io and api-changelog.honeycomb.io disallow unnamed crawlers, so we didn't read them and scored… · Transparency & trust, The hosted server is closed, under published Terms of Service and Honeycomb Intelligence Supplemental Terms.
- Sources: 32, open questions: 8, both in the full twin
- Capabilities: observability.traces, observability.metrics, observability.logs
- JSON: https://www.anchorterminal.com/api/v1/tools/honeycomb-mcp.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/honeycomb-mcp.svg` or a link to https://www.anchorterminal.com/tools/honeycomb-mcp from a page on honeycomb.io or one of its subdomains, or the README of github.com/honeycombio/agent-skill, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.

## Before you call it

1. Use https://mcp.honeycomb.io/mcp for US teams and https://mcp.eu1.honeycomb.io/mcp for EU teams. A team's data lives in one region only
2. If no tools appear after sign-in, ask a Team Owner to enable Honeycomb Intelligence in team settings
3. Send `X-MCP-Toolset: query` or `canvas` to load a subset of the 34 tools. An empty or unknown value makes every request fail
4. Call `get_workspace_context` first, then `find_columns`, before `run_query`. Pass absolute time ranges as epoch timestamps
5. For unattended runs send `Authorization: Bearer <KEY_ID>:<SECRET_KEY>` from a Management key with only Model Context Protocol and Environments scopes

## Connect

```bash
claude mcp add honeycomb --transport http https://mcp.honeycomb.io/mcp
```
Full config and headless snippets are in the full page. Through letme (picks today, calling later): https://letme.dev/honeycomb-mcp

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| Grafana MCP Server | BB | 74.5 | observability.metrics, observability.logs, observability.traces | https://www.anchorterminal.com/tools/grafana-mcp-server.min.md |
| Datadog MCP Server | C | 56.6 | observability.metrics, observability.logs, observability.traces | https://www.anchorterminal.com/tools/datadog-mcp.min.md |
| Render MCP Server | B | 63.6 | observability.logs, observability.metrics | https://www.anchorterminal.com/tools/render-mcp-server.min.md |
| LaunchDarkly | B | 69.2 | observability.logs | https://www.anchorterminal.com/tools/launchdarkly.min.md |
| PostHog | B | 68.4 | observability.logs | https://www.anchorterminal.com/tools/posthog.min.md |


## Panel reviews (0, desk reviews from public material, no calls made)