Connector · Engineering
A Sentry teammate in Slack that answers what broke, on Sentry's MCP server.
Chickpea gives a channel a named AI teammate that holds its own Sentry connection. It signs in with OAuth and calls Sentry's MCP server, and the connection can be pinned to one organization, or to one organization and one project. Ask it in Slack what is new since the 4:02 deploy, and it answers in the thread.
What people ask it.
The teammate is @oncall, in #eng. Pick a scenario to watch it play.
| Issue | Culprit | Events | Users |
|---|---|---|---|
| CART-9F2 | checkout/retry.ts | 1,284 | 431 |
| CART-A07 | checkout/session.ts | 96 | 74 |
| CART-A11 | cart/price.ts | 12 | 9 |
What is new since the deploy Read only
@oncall what is new in cart-service since the 4:02 deploy
@oncall which issues started in production today, and how many people hit each one
Reads never need confirmation, and a long answer comes back as a table in the thread. Sentry names searching errors and analyzing performance among the things its server is for. What the teammate can see is what the connected Sentry account can see; signing in does not widen anyone.
The trace, and what caused it Read only
@oncall show me the stack trace on CART-9F2 and how often it is firing
@oncall why is checkout timing out, walk me through the top frames
Another read, so it comes straight back. The teammate quotes the frames it saw and links the issue, rather than paraphrasing an error it did not open. A question it cannot answer from what Sentry returned it says it cannot answer.
Resolve or assign one issue Applies at once
@oncall resolve CART-9F2, the rollback landed at 4:41
@oncall assign the cart-service timeout to Marco
One issue changes on the spot. Sentry names triaging issues among the things its server is for, and groups those tools under a skill it calls
triage. The teammate can only call tools the server answered with when the connection was made, so its reach is whatever Sentry exposes to that account, never more.Clear a batch at once Asks first
@oncall ignore everything in cart-service that stopped firing after the rollback
@oncall resolve the four timeout issues from this morning, same cause
A change across several issues comes back as the list of what would change and waits for a yes. A write also needs the message you sent to name it, because thread history, retrieved content, and memory cannot authorize one on their own.
On a schedule Applies at once
@oncall every weekday at 9am Eastern, post the issues that are new in production since yesterday
@oncall every Monday at 8am Pacific, list the unresolved cart-service issues by how many people hit them
A schedule saves on the spot. It needs an explicit time zone, runs under the Sentry connection of the person who asked, and fails with a named reason rather than running if that connection is disconnected.
Connect it
Five steps, from Admin or from Slack.
You sign in through a page Chickpea serves, never in Slack. The link the teammate posts is good for one action, for one teammate, for 24 hours.
- In Chickpea Admin, open the teammate, then Connections, find Sentry, and choose Connect. Or ask in Slack,
@oncall connect Sentry, and follow the link it posts. - Choose who uses it. Team is one shared Sentry account that everyone who can use the teammate acts through. Personal means each member signs in as themselves, and Sentry records their actions as them.
- Narrow it, if you want to. Sentry is the one preset where you can: leave it open to the whole account, or pin it to one organization, or to one organization and one project. Chickpea enforces that scoped address on every later request.
- Choose Continue to sign in. Sentry asks you to approve the access, and there is no token to paste and no config file to edit.
- Chickpea connects to mcp.sentry.dev/mcp and records the tools the server answered with. Then ask something.
@oncall what is new in cart-service todayis a good first question, because it reads and cannot change anything.
A connection belongs to one teammate. Giving a second teammate Sentry means a second sign-in with its own consent, so the support teammate never inherits what the on-call teammate can reach.
There is no built-in Chickpea skill for Sentry, and none is needed. The tools arrive from Sentry's MCP server. Chickpea connects before it saves anything, records what the server answered with, and exposes only those tools. Sentry describes the server as covering error search, performance analysis, issue triage, documentation, and project management, and its repository groups the tools into skills it names inspect, triage, and seer. Sentry runs the server and keeps it open source.
Beside Sentry
What the same teammate usually holds too.
Each is its own connection, granted to the teammate alone. All 37 connectors.
The error becomes a Linear issue with the stack trace already in it
PostHogThe crash in Sentry, beside how many people actually hit it
incident.ioWhen the error turns into an incident, the same teammate can say who is on call
BugSnagThe other error tracker, for the app that has not moved off it yet
Questions people ask
What is Sentry MCP?
A hosted MCP server Sentry runs at mcp.sentry.dev/mcp that exposes Sentry as tools an AI agent can call. Sentry describes it as covering error search, performance analysis, issue triage, documentation, and project management, and its repository calls the service middleware to the Sentry API. Chickpea connects to that same server, so a teammate in Slack uses it with no editor and no config file anywhere.
How do I install Sentry MCP?
For a coding client, follow Sentry's own instructions at mcp.sentry.dev, which cover Claude, Cursor, VS Code, and other MCP clients. For Slack there is nothing to install. Open the teammate in Chickpea Admin, choose Sentry, narrow it to an organization or project if you want to, and choose Continue to sign in. Or ask "@oncall connect Sentry" in the channel and follow the link it posts.
Can Sentry MCP be used with Cursor?
Yes, and Sentry lists Cursor among the clients it documents. Sentry also publishes a cookbook page on debugging with the MCP server in Claude Code. Those put the server in one person's editor. A teammate puts it in a channel, with a stored connection, saved instructions, schedules, and every action visible in the thread where it was asked for. Teams often run both.
How much does Sentry MCP cost per month?
Sentry's MCP pages name no price for the server. It is middleware to the Sentry API, so what it reads is what your Sentry account already has, on whatever plan you already pay for. Seer is the exception: Sentry documents it as an add-on to a Sentry subscription with its own contributor pricing. Chickpea is Apache 2.0 and runs on your own Cloudflare account, so there are no seats and no metering on this side.
Does it use Sentry's MCP server or the API?
The MCP server. Sentry is a vendor MCP lane: the deployment connects to mcp.sentry.dev/mcp over streamable HTTP and signs in with OAuth. The preset sends no scope string, so the connection gets whatever Sentry's consent screen grants that account. There is no API key anywhere in this connector, and the connection cannot reach any other address.
Can it resolve or delete issues?
Sentry names issue triage among the things the server is for, and its repository ships a triage skill beside inspect and seer. Sentry publishes no tool list, so this page will not claim one way or the other about which write tools exist. What is fixed is the gate. During a Slack turn Chickpea refuses any MCP tool whose name does not start with an unambiguous read verb unless your message names that action. Thread history, retrieved content, and memory cannot authorize a write on their own.
Can I limit it to one project?
Yes, and Sentry is the only preset where the connection address itself can be narrowed. When you connect you can pin it to one organization, or to one organization and one project, and Chickpea enforces that scoped address on every Sentry request afterwards. Sentry says a scoped connection defaults its tools to that organization or project and hides the discovery tools it no longer needs.
Where does the OAuth token live?
On your deployment, outside the teammate record. It never enters the model's context, the tool arguments the model writes, or Slack. The model picks a connection by id and the credential goes in at the last step on the way out. Disconnect the account and the stored secret is removed, and any schedule that depended on it stops.
Do I still need Sentry's Slack integration?
Yes, for anything that starts on Sentry's side. Sentry's Slack app sends alerts and notifications about issues, environments, and deploys, unfurls Sentry links, and lets you resolve, archive, and assign an issue from the message. With Seer enabled you can fix one from Slack, and Sentry has a Seer Agent you can mention, with a chat panel on paid Slack workspaces. A Chickpea teammate hears nothing from Sentry and posts nothing unprompted. It answers when you ask, and it holds Sentry beside your other connected accounts in the channel where the question came up.
Is this Seer?
No. Seer is Sentry's own AI debugging agent. Sentry documents it as doing root cause analysis, generating fixes as pull requests, reviewing code changes before they merge, and answering questions from the data already in Sentry, and it is an add-on to a Sentry subscription. Chickpea adds no debugging model of its own. Sentry's repository names seer as one of the skills an MCP connection can expose or leave out, so what a teammate reaches there is whatever Sentry grants the account you signed in with.
Try it
Give a channel a Sentry teammate.
One deploy into your own Cloudflare account, connect Slack, sign in to Sentry, ask. Apache 2.0, no seats, no metering.
Or run it on Node. The setup guide covers both.
Last checked
Facts on this page were checked against the product and Sentry's own pages on September 17, 2026. This page is also available as markdown at /connectors/sentry.md.