# Honeycomb MCP connector The Honeycomb connector lets Claude, ChatGPT, Cursor, or the Elaichi Agent run queries, read traces, and manage boards and triggers in Honeycomb, with each person signed in as themselves and every call logged. Source: https://elaichi.ai/connectors/honeycomb/ ## Facts | | | | --- | --- | | Application | Honeycomb | | Category | Observability | | AI tools | 34 | | Authentication | Connects over OAuth | | Bring your own OAuth app | No | | Native MCP | Yes. Honeycomb builds and runs this MCP server. Elaichi adds sign-in, access controls and an audit log on top | | Support for its tools | www.honeycomb.io/support | | MCP endpoint | https://api.elaichi.ai/mcp | | Works with | Claude, ChatGPT, Cursor, any MCP client, and the Elaichi Agent | | Tools advertised by name | No. Connected tools are never listed one by one, however few there are. The endpoint advertises `search_tools` and `execute_tool` instead | ## What you can ask once Honeycomb is connected - Which spans made checkout slow in the last hour? - Run BubbleUp on errors in the api-gateway dataset today - Which SLOs are burning budget fastest this week? ## Connect Honeycomb in Elaichi This happens once for the organization, before any client is involved. 1. Open Connections, choose Add connection, and pick Honeycomb. 2. Optionally set Share with, then press Connect. 3. Approve it in Honeycomb. Honeycomb's own window opens. Whoever approves it decides what this connection can reach. Credentials are vaulted and nobody, including the AI, reads them back. The connection becomes a toolbox immediately, so you can curate which Honeycomb tools are exposed, rename them, or freeze arguments before anyone points a client at it. ## Honeycomb MCP connector for Claude Endpoint: https://api.elaichi.ai/mcp 1. Open Customize, then Connectors. 2. Press Add. 3. Name it, paste the MCP server URL, then Continue. 4. Sign in and approve. On Team and Enterprise, an Owner adds it once. Everyone else turns it on for themselves. ## Honeycomb MCP connector for ChatGPT Endpoint: https://api.elaichi.ai/mcp 1. Open Plugins, then press the + button. 2. Name it and paste the endpoint into Server URL. 3. Leave Authentication on OAuth, then tick the risk acknowledgement. 4. Press Create, then sign in and approve. Works on the web today. The plugin directory lives at chatgpt.com/plugins. ## Honeycomb MCP connector for Cursor Endpoint: https://api.elaichi.ai/mcp 1. Open `~/.cursor/mcp.json`. 2. Add the endpoint under `mcpServers`. 3. Reload Cursor, then sign in and approve. Set up per machine, so repeat it on each computer you work from. ## Connect Honeycomb to any MCP client Endpoint: https://api.elaichi.ai/mcp 1. Add the endpoint as a remote MCP server. 2. Sign in and approve. The Elaichi Agent already has these tools, with nothing to set up. ## What the consent screen decides Only Read is granted by default, which is not enough to call a Honeycomb tool. Over MCP there is no trusted place to confirm a write in the moment, so the consent screen is the standing approval rather than a formality. Grant Read and Run tools. Think hard before granting Delete, which reaches into connected apps and cannot be undone. ## What teams do with Honeycomb through Elaichi ### Find the slow request behind a page On-call. Ask for the trace behind a latency alert, see which spans took the time, and get a plain summary before you open a laptop. ### Explain a spike with BubbleUp SRE. Run BubbleUp on the error rate in a dataset and have the agent tell you which build, region, or customer stands out from the rest. ### Check a service before a deploy Engineering. Pull the service map and the anomaly profiles for the service you are about to ship, and see what depends on it. ### Keep triggers and boards tidy Platform. Create a trigger for a new dataset or update the queries on a board without clicking through the Honeycomb UI. ### Confirm a customer's reported error Support. Ask for recent spans from that customer's requests and get a plain answer you can paste into the ticket. ### Report on SLOs for the weekly review Product. List the SLOs and the queries behind them, then turn them into a short status the whole team can read. ## Elaichi vs Zapier MCP vs Composio for Honeycomb All three can connect Honeycomb to an AI assistant, and all three have admin controls. They differ in where access lives and how you pay. | What to check | Elaichi | Zapier MCP | Composio | | --- | --- | --- | --- | | Where the AI connects | One address for the whole organization. Endpoint: https://api.elaichi.ai/mcp | A server per member, created at sign-in. | An MCP endpoint per team, or an SDK. | | Control over Honeycomb tools | Allow or restrict single Honeycomb tools, per role or user. | App and action restrictions on the account. | Role permissions, down to the action. | | Record of calls | One audit entry per Honeycomb call. | A History tab of tool calls. | A log of every tool call. | | Single sign-on | SAML or OIDC, plus SCIM, on Gold. | SAML on Enterprise. | SAML and OIDC on Enterprise. | | Price | $15 per user per month. | 2 tasks per successful call. | Billed per tool call. | Sources: Zapier MCP [docs](https://docs.zapier.com/mcp/get-started/quickstart), [security](https://docs.zapier.com/mcp/manage/security), [usage](https://docs.zapier.com/mcp/features/usage); Composio [docs](https://docs.composio.dev/docs/composio-connect), [gateway](https://composio.dev/mcp-gateway), [enterprise](https://composio.dev/enterprise), [pricing](https://composio.dev/pricing). Checked September 2026. Longer take: [Zapier MCP alternative](/blog/zapier-mcp-alternative/) and [when you don't need an MCP gateway](/blog/when-you-dont-need-an-mcp-gateway/). ## Frequently asked questions ### How do I connect Honeycomb to Claude? Connect Honeycomb in Elaichi first: you sign in to Honeycomb over OAuth and approve access, with no OAuth application to register and no client ID or secret to generate. Then open Claude, go to Customize, then Connectors, then Add, and paste https://api.elaichi.ai/mcp. Claude signs you in to Elaichi and Honeycomb is available from that point. ### Does Honeycomb work with ChatGPT and Cursor as well as Claude? Yes. Honeycomb is connected once in Elaichi and the same endpoint, https://api.elaichi.ai/mcp, works in Claude, ChatGPT, Cursor, any other MCP client, and the Elaichi Agent. There is nothing extra to set up per client. ### What can an AI agent actually do with my Honeycomb data? It can run queries against your Honeycomb datasets, read the results, pull a trace and walk its spans, run BubbleUp to explain an outlier, and look at the service map and anomaly profiles for a service. It can also list boards, triggers, and SLOs, and create or update boards and triggers. Short, specific asks such as which spans were slow in checkout this hour work better than long paragraphs. ### Does connecting Honeycomb give the AI access to every environment and dataset? No. Access follows the person who signed in to Honeycomb, so the agent sees the environments and datasets that person can already see and nothing more. Elaichi can narrow that further for a team or a toolbox, but it can never widen it beyond what Honeycomb itself allows. ### Can my team share one Honeycomb connection? Yes. One person connects Honeycomb in Elaichi and shares it with a team, and nobody else ever handles a key or a credential. Each teammate still signs in to Elaichi as themselves, so every Honeycomb query and trace lookup in the audit log is tied to the person who made it. ### Can I stop an agent from creating or changing triggers and boards in Honeycomb? Yes. Restrictions in Elaichi apply per action, so you can allow running queries and reading traces in Honeycomb while blocking the creation or updating of triggers and boards. A restricted action is never advertised to the AI client, so no prompt can reach it. ### What happens to a Honeycomb connection when someone leaves? Offboarding that person in Elaichi ends their Honeycomb access at once, across every client they used. A shared Honeycomb connection keeps working for everyone else on the team. If you want it gone entirely, disconnecting Honeycomb once in Elaichi removes it from Claude, ChatGPT, Cursor, and every other client at the same time. ### Does the Honeycomb MCP connector work with Gemini, Codex, Claude Code or other MCP clients? Yes. Honeycomb is reached over the same MCP endpoint every client uses, so anything that speaks MCP can call it — Gemini, Codex, Claude Code, Windsurf, Cline, Zed and OpenCode among them — alongside Claude, ChatGPT, Cursor, and the Elaichi Agent. The tools on offer and the access behind them are identical whichever client asks. Only the setup screen differs. ### Is Elaichi an alternative to Zapier MCP for Honeycomb? Yes. Both let Claude, ChatGPT or Cursor use Honeycomb. Zapier MCP fits a team that already automates in Zapier, since each person signs in and acts as themselves in that account. Elaichi fits when IT wants one address for the whole company, per-tool rules by role, and a record of every Honeycomb call. ### How is Elaichi different from Composio for Honeycomb? Composio gives AI agents tools and sign-in handling across 1,000+ apps, for developers building agents or people using an assistant, billed per tool call. Elaichi gives a company's own people governed access to Honeycomb: one address, restrictions per role or user, and $15 per user per month. Both have role permissions and a log of every call. ## All 34 Honeycomb tools Every tool below is callable through https://api.elaichi.ai/mcp once Honeycomb is connected, subject to the toolbox it is in and the restrictions on the caller. - **Get workspace context** (Get). Returns the team name, current time, and a list of environments with their slugs and dataset counts. Takes no parameters. Agents call this tool first to orient themselves before doing anything else. Most prompt-driven workflows start here so the agent knows which environments… - **Get environment** (Get). Returns details for a single environment, including its datasets and calculated fields, sorted by most recent activity. Returns up to 100 datasets per call. - **Get dataset** (Get). Returns metadata and the full column schema for a single dataset, including columns and calculated fields in a unified list sorted by last write time. Returns up to 100 columns per call by default. - **Get dataset columns** (Get). Returns the full column schema for a single dataset, with optional sample values for specific columns. For metrics datasets, returns metric names by default. Pass a `metric_name` to discover the attributes available for filtering or grouping a specific metric. - **Run query** (Search). Runs a time-series aggregation query against a Honeycomb dataset and returns computed results. Supports compound queries (query math), per-calculation filters, formulas, breakdowns, calculated fields, and relational trace prefixes (`root.`, `parent.`, `child.`, `any.`). Agents… - **Get query results** (Search). Retrieves results and metadata from a previously executed query run. Accepts a Honeycomb query URL, a query run primary key, or a query ID (which returns the most recent run). Agents use this tool to fetch the output of a saved or earlier query without re-running it. - **Find queries** (Action). Searches query history and saved queries by intent and returns matching queries with their run primary keys. Pair the result with `get_query_results` to fetch the actual data. Useful when you want the agent to learn from prior investigations or reuse a saved query. - **Find columns** (Action). Searches for columns and calculated fields by intent across one or all datasets in an environment. Agents use this tool to find relevant fields based on natural-language keywords rather than exact column names. Honeycomb's Weaver registry feeds richer descriptions into this… - **Run bubbleup** (Run). Runs a BubbleUp analysis on an existing query result to identify what makes a selected subset of data different from the baseline. BubbleUp compares value distributions across columns and surfaces the statistically significant differences. - **Get trace** (Get). Retrieves all spans for a specific trace ID and renders them as a waterfall. Agents can use the parameters for this tool to zoom in to specific subtrees of a trace and identify errors. - **List spans** (List). Lists span names in trace data, ranked by count, with how often each is a trace root and which dataset the count came from. - **Get span details** (Get). Returns a summary of attributes and their common values observed on spans with a specific name, including which attributes are populated, how many distinct values each has, and the top observed values. - **Get service map** (Get). Returns a snapshot of service-to-service call dependencies for a specified time range. Powers the same graph as the Service Map view in Honeycomb. - **Get anomaly service profiles** (Get). Returns anomaly detection service profiles for the current team. Requires that anomaly detection be enabled for the team. - **Search semconv** (Search). Searches the semantic convention registry for attributes matching a query. - **Get semconv attribute** (Get). Returns full definitions of one or more semantic convention attributes by their exact names. - **List semconv namespaces** (List). Lists the top-level semantic convention namespaces available in the team's registry, including any team-specific attribute customizations. - **List boards** (List). Lists Boards in an environment, or returns the full contents of a single Board by ID. Use this before `update_board` to inspect existing panels and IDs. - **Create board** (Create). Creates a Board with query panels, Service Level Objective (SLO) panels, and text (Markdown) panels. Panels appear in the order you specify them and can include explicit width and height. Supports preset filters that become filter dropdowns on the Board. To learn more about… - **Update board** (Update). Updates an existing Board. Supports renaming, adding, removing, updating, and reordering panels, plus replacing preset filters and tags. - **Get triggers** (Get). Lists Triggers for the team, or returns detailed configuration for a single Trigger, including recipients with their type, name or target, and ID. - **Create trigger** (Create). Creates a Trigger that fires alerts when query results cross a threshold. To learn more about Triggers, visit Create a Trigger. - **Update trigger** (Update). Updates an existing Trigger. - **Get slos** (Get). Lists SLOs for the team, or returns detailed status and graphs for a single SLO. - **Create slo** (Create). Creates an SLO with an auto-created Service Level Indicator (SLI) derived column. Provide the SLI expression and an alias, and the tool creates the derived column if needed, validates the expression, and then creates the SLO. To learn more about SLOs, visit Create an SLO. - **Update slo** (Update). Updates an existing SLO. Partial-update semantics apply; omitted fields keep their current values. Dataset associations are fixed at creation time; create a new SLO to use different datasets. - **List recipients** (List). Lists all pre-registered notification recipients for the team, including email, Slack, PagerDuty, and webhooks, with their IDs. - **Create recipient** (Create). Creates a notification recipient that can be attached to Triggers and SLO burn alerts. Returns the recipient ID for use with `create_trigger` and `update_trigger`. To learn more about recipients, visit Recipients for Notifications. - **Canvas agent invoke** (Action). Sends a message to the Canvas agent. If an investigation ID is provided, the message is routed to the user's agent in that investigation. Otherwise, a new investigation is created. - **Canvas agent poll response** (Action). Polls for the result of a previously-issued `canvas_agent_invoke` call. To learn more about Canvas investigations, visit Canvas. - **List aiconversations** (List). Lists recent AI agent conversations (`gen_ai.conversation.id` values) in an environment, ordered by total event count. Includes a per-conversation breakdown of agents, services, event counts, error counts, and token usage. - **Get aiconversation** (Get). Returns the full event timeline for a single AI conversation by its `gen_ai.conversation.id` value, including every LLM call, tool call, and related agent event with span name, operation, agent name, model, tool name, duration, and error detail. Also includes an aggregate… - **Refinery docs** (Action). Reads from Honeycomb Refinery documentation, so your agent can answer Refinery configuration questions without leaving the conversation. - **Feedback** (Action). Submits feedback about the Honeycomb MCP server experience to the Honeycomb team that maintains it. Ask your agent to "submit feedback to Honeycomb" with your message.