A Sentry MCP for automations: errors become tasks, releases annotate the work
Connect one auth token and Sentry issues, events, releases, deploys, projects, and teams become workflow steps for visual automations and AI agents. Sentry already speaks MCP for chat; this connector is for the step after the conversation, where a sweep over unresolved errors opens tasks and a deploy annotates the work that shipped it.
What the Sentry MCP connector does
The Sentry connector for Mattering Automations is 33 nodes across the Web API v0: organizations, projects, and teams, the issue surface (search with Sentry's structured query syntax, single and bulk updates, tag breakdowns, and tightly guarded deletes), the error events grouped under each issue, releases and deploys, project client keys, organization event counts, and read-only audits of issue and metric alert rules.
Every node speaks MCP, the open protocol for connecting AI assistants to tools, so Claude, ChatGPT, and any MCP-capable assistant can use them, alongside Matter, the AI assistant built into the platform. Sentry ships an official MCP server of its own, and for pure chat access to Sentry it is a fine place to start. This connector exists for what comes next: the same 33 nodes are steps in visual automations, so what an assistant can look up, a workflow can act on, on a schedule, next to the rest of the work.
Because Mattering is a work platform rather than middleware, the natural payoff is cross-app: a scheduled sweep over unresolved issues can open tasks in Mattering Productivity with the short id, permalink, and counts attached, a daily digest of event volumes can land in a Mattering database, and a release created from the canvas annotates the work that shipped it. Sentry: Update Issue is the workhorse in these flows: resolve, ignore, assign, and set priority, with every change reversible by setting the field back.
The destructive surface is exactly two nodes, and both are fenced. Sentry: Delete Issue and Sentry: Bulk Delete Issues queue permanent, asynchronous removal of an issue and all its events with no undo, and the bulk node requires an explicit list of issue ids: the raw API also accepts a filter-wide delete when ids are omitted, and that project-wide wipe is deliberately not exposed on the canvas. Everything else is a read (24 of the 33 nodes) or a reversible write.
One connection, three surfaces
- Visual automations. Drag Sentry nodes onto the canvas and wire them to triggers, conditions, and other apps.
- AI agents, yours or ours. These nodes speak MCP, the open protocol, so Claude, ChatGPT, and any MCP-capable assistant can use them, and Matter, the agent built into the platform, uses them natively.
- The whole platform. Combine with native tasks, projects, and other connectors, no middleware in between.
Every Sentry node, ready to drop on the canvas
33 nodes, generated from the connector itself so this list always matches what ships. Expand a category to browse.
Read and query nodes 24
Sentry: Get Issue
Retrieve one issue's full details: status, substatus, priority, assignee, counts, first and last seen. Use Sentry: Resolve Short ID first if you only have the BACKEND-API-1A form. Requires event:read.
Sentry: Get Issue Event
Retrieve one event of an issue, with full detail including tags and entries. event_id accepts the special values latest (the default here), oldest, and recommended as well as a specific 32-character event ID, so with an empty event_id this node returns the issue's latest event, the standard "show me the newest occurrence" step. Requires event:read.
Sentry: Get Organization
Retrieve one organization's full details: settings, feature flags, access, and (unless detailed is 0) its teams and projects. Requires org:read.
Sentry: Get Organization Event Counts
Retrieve event counts for an organization (the stats_v2 endpoint): time series of accepted, filtered, and dropped events, groupable by outcome, category, reason, or project. The quota and volume view, for usage digests and spike alerts. Requires org:read.
Sentry: Get Project
Retrieve one project's full details: platform, teams, features, and settings such as digest delays and resolve age. Requires project:read.
Sentry: Get Project Event
Retrieve one event by its 32-character hexadecimal event ID (as reported by the SDK client), with full detail. Use this when a workflow starts from an event ID, for example out of a webhook or a log line. Requires project:read.
Sentry: Get Release
Retrieve one release with commit and deploy counts, optionally including release health data (crash-free sessions and users) when health is true. Requires project:releases or project:read.
Sentry: Get Team
Retrieve one team's details, including member count and the acting token's access. Requires team:read.
Sentry: List Issue Alert Rules
List a project's issue alert rules: conditions, actions, and frequency for the when-an-issue-matches alerts. Read-only auditing surface; rule mutation is deliberately not exposed here. Sentry's docs mark this endpoint deprecated in favor of the newer alerts APIs, but it remains live and is the stable list form. Requires alerts:read or project:read.
Sentry: List Issue Events
List the individual error events grouped under one issue. By default returns event summaries; set full to include the complete event body with stacktrace (much larger output, page accordingly). Requires event:read.
Sentry: List Issue Tag Values
List the values of one tag across an issue's events, with per-value counts: which browsers, releases, URLs, or users an issue actually hits. Requires event:read.
Sentry: List Metric Alert Rules
List an organization's metric alert rules: the threshold-based alerts over error counts, sessions, and performance metrics, with their triggers and thresholds. Read-only auditing surface. Requires alerts:read or org:read.
Sentry: List Organization Issues
List issues across a whole organization, filterable to specific projects and environments, with the same structured search syntax as the project-level list. The org-wide triage view for daily digests and escalation sweeps. Requires event:read.
Sentry: List Organization Members
List an organization's members with email, name, org role, and invite status. The member id returned here is what member-management endpoints take; the assignedTo field on issue updates takes the USER's id, username, or email instead. Requires member:read.
Sentry: List Organizations
List the Sentry organizations the auth token can see, with slug, name, and status for each. Needs no parameters at all, which makes it both the credential health check and the natural first step of any Sentry workflow: every other node needs the organization slug this returns. A user token lists the user's organizations; an internal-integration token lists its own organization. Requires the org:read scope.
Sentry: List Project Client Keys
List a project's client keys (DSNs), the ingestion endpoints SDKs send events to. Each key carries its public DSN and rate limit settings. DSNs authorize event INGESTION only, not API reads, but treat the output with care anyway. Requires project:read.
Sentry: List Project Events
List a project's error events across all its issues, newest first, optionally windowed by time. Event streams for audit trails and export flows; for triage, the issue lists are usually the better unit. Requires project:read.
Sentry: List Project Issues
List a project's issues (error groups) with Sentry's structured search syntax. The default query is is:unresolved; combine filters like is:unresolved level:error release:backend@2.4.1 assigned:me. Returns issue id, short id, title, status, and counts; the issue id feeds the get, update, delete, and event nodes. Requires event:read.
Sentry: List Projects
List an organization's projects with slug, platform, and team. The project slug returned here is what the project-scoped issue, event, key, and alert nodes take. Requires org:read.
Sentry: List Release Deploys
List a release's deploys: which environments it went to and when. Requires project:releases or project:read.
Sentry: List Releases
List an organization's releases, newest first, filterable by project and environment. query is a case-insensitive substring match on the version. Requires project:releases or project:read.
Sentry: List Team Projects
List the projects a team owns. Useful before Sentry: Create Project to see what already exists under the team. Requires project:read.
Sentry: List Teams
List an organization's teams. The team slug returned here feeds Sentry: Create Project and team:<id> issue assignment. Requires team:read (org:read also grants it).
Sentry: Resolve Short ID
Translate a human-facing short ID like BACKEND-API-1A (the form Sentry shows in the UI, commit messages, and chat links) into the numeric issue ID and project slug that the issue and event nodes need. Requires event:read.
Action nodes 9
Sentry: Bulk Delete Issues
DESTRUCTIVE: queue a list of a project's issues for permanent deletion, groups and events both, with no undo. This node REQUIRES an explicit id list on purpose: the raw API deletes by status filter when ids are omitted, and that project-wide wipe is deliberately not exposed on the canvas. Requires event:admin.
Sentry: Bulk Update Issues
Mutate several of a project's issues in one call: pass issue IDs and the fields to set (typically status: resolved to close out a list). CAUTION: when the id list is omitted the API applies the change to the project's issues matching the status filter (up to its internal bulk limit), so always set ids unless a sweeping change is intended. IDs outside the project are silently ignored, per the docs. Field changes are reversible; the merge flag is not cleanly so (unmerging is manual). Requires event:write.
Sentry: Create Deploy
Record a deploy of a release to an environment, which timestamps the release for that environment and triggers deploy notifications. Note: deploys cannot be deleted via the API, so a test deploy on a scratch release leaves a permanent (harmless) record. Requires project:releases (or org:ci), so this also works with an organization token.
Sentry: Create Project
Create a new project under a team. Note the team-scoped path: Sentry creates projects via a team, not directly on the organization. Returns the created project object; list its DSN with Sentry: List Project Client Keys to start sending events. Requires project:write.
Sentry: Create Release
Create a release for one or more projects, so events tag themselves to it and regressions can say which version reintroduced them. Idempotent-ish: creating an existing version again updates its project associations rather than failing. Follow with Sentry: Create Deploy to record where it shipped. This is the surface Sentry's own org tokens are scoped for, so it works with all three token types. Requires project:releases (or org:ci).
Sentry: Delete Issue
DESTRUCTIVE: queue one issue for permanent deletion, removing the group and all its events. The API answers 202 Accepted and deletes asynchronously; there is no undo. Prefer Sentry: Update Issue with status resolved or ignored unless the data itself must go. Requires event:admin.
Sentry: Update Issue
Update one issue: resolve, unresolve, or ignore it, assign it to a user or team, set its priority, bookmark or subscribe. Only the fields you set are changed, and every change here is reversible by setting the field back. The workhorse of Sentry triage automations. The raw API also accepts merge and discard flags via extra_body; discard deletes future events for the issue, so treat it as destructive if used. Requires event:write.
Sentry: Update Project
Update a project's settings: rename, change the slug or platform, bookmark it, set the mail subject prefix, or the auto-resolve age in hours. Only the fields you set are changed. Requires project:write for most fields. Careful with slug changes: existing URLs and configs referencing the old slug break.
Sentry: Update Release
Update an existing release's metadata: ref, URL, release date, or attached commits. The version itself cannot be renamed. Requires project:releases.
Connecting Sentry takes one credential
Sentry issues User Auth Tokens from Settings > Account > API > User Auth Tokens, with scopes you choose at creation. An internal integration token, created under an organization's Developer Settings, also gives full coverage and acts as the organization rather than one person, which makes it the better choice for a shared automation credential. Either kind is sent as a Bearer credential on every request. Avoid Organization Tokens: their permissions are fixed for CI release tasks, so most read nodes return 403 with one.
In Sentry, open Settings > Account > API > User Auth Tokens and create a token with the scopes your automations need: org:read, project:read, team:read, event:read, and event:write cover triage; add project:write, member:read, and project:releases for the full surface, and event:admin only if you will use the delete nodes.
In Mattering Automations, open Connectors, click Add Connector, choose Sentry, and create a credential with that token.
Click Test Connection. Mattering calls Sentry's organizations endpoint, a zero-parameter read that confirms the token authenticates.
All 33 Sentry nodes appear in the canvas palette. Run Sentry: List Organizations first: it returns the organization slug every other node needs.
Common uses for the Sentry MCP
Triage that happens where the work is
Sentry: List Project Issues and Sentry: List Organization Issues search with Sentry's structured query syntax (the default is is:unresolved), and Sentry: Update Issue resolves, ignores, assigns, and prioritizes what they find. Every typed field on the update node is reversible by setting it back, which is what makes triage safe to automate.
Errors become tasks automatically
A scheduled sweep over unresolved issues opens a task in Mattering Productivity for each new one, with the short id, title, and event counts attached. When the task closes, Sentry: Update Issue marks the issue resolved, so the tracker and the task board stop drifting apart.
Releases and deploys from the pipeline
Sentry: Create Release tags events to a version so regressions can say which release reintroduced them, and Sentry: Create Deploy timestamps where it shipped and triggers deploy notifications. These are the nodes CI naturally drives, and they are the one surface even a CI-scoped Organization Token can reach.
An agent that answers what broke
Give an assistant Sentry: Resolve Short ID to turn the BACKEND-API-1A form from a commit message or chat link into a real issue id, Sentry: Get Issue for status and counts, Sentry: Get Issue Event for the latest occurrence with its stacktrace, and Sentry: List Issue Tag Values to say which browsers, releases, or users an issue actually hits.
Volume digests and spike awareness
Sentry: Get Organization Event Counts returns time series of accepted, filtered, and dropped events, groupable by outcome, category, or project. On a schedule it becomes a usage digest in a Mattering database, or a message to the team when volume spikes past a threshold.
Alert rules audited, not guessed
Sentry: List Issue Alert Rules and Sentry: List Metric Alert Rules read out the conditions, actions, and thresholds actually configured, so a periodic audit can flag projects with no alerting at all. Rule mutation is deliberately not exposed: misconfigured alerting silently loses pages.
Example: Unresolved errors become tasks, and closing the task resolves the error
Built on the Automations canvas in a few minutes, or described in plain language to Matter and built for you.
Trigger: a schedule runs the sweep every morning, or a Sentry alert rule pointed at a Mattering webhook trigger starts it the moment something fires.
Sentry: List Organizations supplies the organization slug that every other node needs.
Sentry: List Organization Issues searches unresolved issues across the projects the team owns, using Sentry's query syntax to keep noise out.
A condition filters to issues first seen since the last run, so the same error never opens two tasks.
Each new issue becomes a task in Mattering Productivity with the short id, title, permalink, and event count, and Sentry: Update Issue assigns the issue to the on-call engineer at the same time.
When the task is completed, Sentry: Update Issue sets the issue to resolved, closing the loop without anyone opening the Sentry dashboard.
Sentry MCP questions, answered
What Sentry operations are supported?
33 nodes on the Web API v0: organizations (list, get, members, event counts, short id resolution), projects (list, get, create, update, client keys), teams (list, get, team projects), issues (project and organization search, get, update, bulk update, tag values, plus two guarded delete nodes), events (per issue and per project, list and get), releases (list, get, create, update, deploys, create deploy), and read-only lists of issue and metric alert rules. Deliberately not shipped: alert rule mutation, organization member and team administration, project deletion and client key mutation, release file uploads, and analytics query surfaces like session health and Discover.
Does this work with Claude, ChatGPT, or other AI assistants?
Yes. Mattering exposes these nodes over MCP, an open protocol for connecting AI assistants to tools, so any MCP-capable assistant, including Claude and ChatGPT, can connect to your workspace and use them. Matter, the assistant built into the platform, uses the same nodes natively.
Sentry ships an official MCP server. Why use this one?
Use whichever fits the job, or both. Sentry's official server gives assistants Sentry access in chat. This connector's 33 nodes are workflow steps first: they run in visual automations on a schedule or behind a webhook, chain into tasks in Mattering Productivity and records in Mattering databases, and are exposed over MCP as well, so Claude, ChatGPT, and Matter get the same tools your workflows use. If the goal is asking about errors, either works; if the goal is a pipeline where errors become tasks and releases annotate work, that is what this one is built for.
Which Sentry token type should I use?
A User Auth Token (Settings > Account > API > User Auth Tokens) with the scopes you pick gives full coverage, but it acts as you and stops working if you leave the organization. An internal integration token, created in the organization's Developer Settings, also gives full coverage, acts as the organization, and its scopes can be edited after creation, which makes it the better shared credential. An Organization Token does not work well here: its permissions are fixed for CI release tasks, so the release and deploy nodes respond while issue, event, member, and stats reads return 403.
Can this connector destroy data?
Only through two clearly marked nodes. Sentry: Delete Issue and Sentry: Bulk Delete Issues queue permanent removal of an issue group and all its events; the API accepts the request and deletes asynchronously, and there is no undo. The bulk node requires an explicit list of issue ids on purpose: the raw API also deletes by status filter when ids are omitted, and that project-wide wipe is deliberately not exposed on the canvas. Both need the event:admin scope, so a token without it cannot delete at all. For everyday cleanup, Sentry: Update Issue with status resolved or ignored is reversible and almost always the right call.
How does pagination work on the list nodes?
Sentry sends cursors in HTTP Link response headers rather than the response body, so the list nodes parse them for you and return a pagination object with next_cursor, prev_cursor, and has_more alongside the data. To fetch the next page, feed pagination.next_cursor into the node's cursor field with otherwise identical parameters, and stop when has_more is false.
Does it work with EU region or self-hosted Sentry?
Yes. Every node carries an optional host field that defaults to sentry.io. Organizations stored in Sentry's EU region set it to de.sentry.io, and self-hosted Sentry installations set their own domain; the API surface is identical in each case. One note: the connector-level connection test calls sentry.io, so verify an EU or self-hosted credential by running Sentry: List Organizations as a node instead.
Can a Sentry error trigger a workflow?
Not as a trigger node in this connector; there is no inbound Sentry trigger. The simple pattern is polling: run Sentry: List Organization Issues on a schedule and act on what is new. For immediate reaction, Sentry's alert rules can call a webhook you configure in Sentry's own UI, and pointing one at a Mattering webhook trigger starts a workflow the moment an alert fires.
Put Sentry to work inside your workspace
Connect once, then use every node in visual automations and AI agent workflows across the Mattering platform.
All product names, logos, and brands are property of their respective owners and are used for identification purposes only. Use of these names, logos, and brands does not imply endorsement or affiliation. Sentry is a product of its respective owner; Mattering is an independent platform.