Menu
Mattering Automations connector

A Bugsnag MCP that turns crash reports into work

Connect one personal auth token and Bugsnag errors, events, pivots, trends, releases, projects and collaborators become workflow steps. An agent can take an alert, pull the freshest stack trace, see which release and device the error is concentrated in, assign it, comment what it did, and open a task, without anyone opening the Bugsnag dashboard.

35 ready-made nodes One credential to connect Works with Claude, ChatGPT, and Matter

What the Bugsnag MCP connector does

The Bugsnag connector for Mattering Automations is 35 nodes on the Bugsnag Data Access API v2: errors with the full workflow verb set (fix, open, ignore, snooze, assign, override severity, link and unlink issues, discard), single and bulk updates, comments, individual events with complete stack traces and metadata, the latest event on an error in one call, pivots that break an error or a project down by any Event Field, the Event Field vocabulary itself, windowed error and project trends, the 30 day session stability trend, releases, projects, organizations and collaborators. One personal auth token, sent with Bugsnag's literal token prefix rather than Bearer, authenticates all of it.

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. An agent handed an error ID can call Bugsnag: Get Latest Event on Error for the freshest report, ask Bugsnag: List Error Pivot Values which app versions and devices the error actually hit, then use Bugsnag: Update Error to assign it and Bugsnag: Comment on Error to write back what it found, so the reasoning lands on the error rather than in a chat log.

SmartBear publishes a SmartBear MCP Server with a documented BugSnag integration, so a vendor route exists and is worth knowing about. This connector is a different shape rather than a replacement. These nodes live on the Mattering canvas next to every other connector, which means Bugsnag steps can run on a schedule, behind a webhook, or inside a longer workflow that also writes a Productivity task, posts to a channel, waits on a human approval, and stores rows in a Mattering database. The same nodes are what an assistant calls over MCP, so the automation and the agent share one credential, one node inventory, and one set of destructive-operation flags.

The safety modeling is stated plainly. Of the 35 nodes, 23 read and 12 act, and five are flagged destructive: Bugsnag: Delete Error, Bugsnag: Delete All Errors in Project, Bugsnag: Delete Event, Bugsnag: Delete Project, and Bugsnag: Remove Collaborator. Bugsnag documents deleting all errors in a project as an action that cannot be reversed, and it takes no filters, so there is no partial version of it. Two scope notes stated up front rather than discovered later: Bugsnag paginates with the Link response header only, so most listers cannot hand a workflow the next page and the honest advice there is to raise per_page rather than page, with releases the exception because they expose an offset a workflow can walk; and Bugsnag's performance monitoring surface (spans, traces, trace fields) is a separate product surface with its own field vocabulary and is deliberately not in this first wave.

The connector also carries the branding detail that trips people up. SmartBear acquired Bugsnag and the API reference now lives on the SmartBear developer portal, but the current machine-readable spec is still titled "BugSnag - Data Access API" and the API host is still api.bugsnag.com, so nothing here needs renaming. The one thing to get right at setup is which host your organization is on: Bugsnag documents a different base server URL per dashboard, and the credential carries an optional host field for exactly that.

One connection, three surfaces

  • Visual automations. Drag Bugsnag 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 Bugsnag node, ready to drop on the canvas

35 nodes, generated from the connector itself so this list always matches what ships. Expand a category to browse.

Read and query nodes 23

Bugsnag: Get Error

Fetch one error with its class, message, status, severity, occurrence and user counts, assignee, linked issues, grouping reason and release stages. Filters scope the counts and the first and last seen times to a window.

Bugsnag: Get Error Trend

Occurrence counts for one error over a time window, bucketed for a histogram. Set the window with since and before, then choose EITHER buckets_count (split the window into that many equal buckets, up to 50) OR resolution (fixed bucket width of 1m, 5m, 30m, 2h or 12h, up to 2000 buckets). Each row carries its own from and to timestamps and an events_count, so the rows are directly chartable.

Bugsnag: Get Event

Fetch one event in full: exceptions with stack traces, threads, custom metadata, request details, app, device and user context, breadcrumbs and feature flags. Custom metadata keeps whatever casing the reporting library sent, so both snake_case and camelCase keys can appear.

Bugsnag: Get Latest Event on Error

Fetch the most recent occurrence of an error as a full report, in one call and without paging a list first. The usual first step after an alert fires: get the freshest stack trace, request and user context for the error.

Bugsnag: Get Organization

Fetch one organization: name, slug, billing contact emails, creator, and the API URLs for its projects and collaborators.

Bugsnag: Get Project

Fetch one project: name, framework type, notifier API key, release stages, grouping and discard settings, open and for-review error counts, and its dashboard URL.

Bugsnag: Get Project Stability Trend

Session stability for a project over the last 30 days, one point per UTC day for the project's primary release stage: total sessions, sessions that saw an unhandled error, users seen, and users who hit an unhandled error. Session stability is 1 minus unhandled_sessions_count over total_sessions_count. Returns HTTP 204 with no body when the project has no session data yet.

Bugsnag: Get Project Trend

Event counts for a whole project over a time window, bucketed for a histogram. Set the window with since and before, then choose EITHER buckets_count (up to 50 equal buckets) OR resolution (1m, 5m, 30m, 2h or 12h, up to 2000 buckets). Filters narrow which events are counted, so this also answers questions like unhandled errors in production per hour.

Bugsnag: Get Release

Fetch one release: app version, version code or bundle version, build label and tool, builder, source control revision and commit URL, errors introduced and seen, and session and user counts for computing stability.

Bugsnag: List Collaborators

List the people in an organization with their IDs, emails, admin flag, pending-invitation state and per-project roles. Returns a bare array. Search by name or email with q.

Bugsnag: List Error Comments

List the comments on an error with their authors and timestamps. Returns a bare array. Comments are sortable by creation time only.

Bugsnag: List Error Events

List the individual occurrences grouped into one error, newest first. This is how an automation gets from an alert to the actual stack traces, users and devices affected. Returns a bare array. Set full_reports for complete event payloads.

Bugsnag: List Error Pivot Values

Break one error down along a single dimension and get the full ranked list: every value of the chosen Event Field with its event count, its share of the total, and when it was first and last seen. Answers which users, versions or devices an error actually hit. Returns a bare array.

Bugsnag: List Error Pivots

Summarize one error across every dimension Bugsnag tracks: for each Event Field, how many unique values it has and which top values account for the occurrences. This is the answer to which release stages, browsers, devices or users an error is concentrated in. Returns a bare array.

Bugsnag: List Errors

List a project's errors: class, message, context, status, severity, occurrence and affected-user counts, first and last seen times, linked issues, and which release stages they happen in. Returns a bare array, so downstream paths start at data[]. Filter with since, before and the Event Field filters object, e.g. error.status eq open. Set histogram to add a trend sparkline to each error.

Bugsnag: List Event Fields

List the Event Fields a project supports: the display IDs usable as filter keys and pivots, their human names, aliases, which comparison types (eq, ne) they accept, and whether each is a built-in field or a custom filter. Run this first to discover what the filters object on the error, event, pivot and trend nodes will accept for this project.

Bugsnag: List Organizations

List the organizations the authenticated token can see, with their IDs, slugs and creators. Returns a bare array, so downstream paths start at data[]. The cheapest authenticated read in this connector and the credential health check to run first: the organization ID from here feeds the project, collaborator and team nodes.

Bugsnag: List Project Collaborators

List the people who have access to one project. This is the set an error can actually be assigned to, so it is the right list to pick from before Bugsnag: Update Error with the assign operation.

Bugsnag: List Project Events

List individual event occurrences across a whole project, newest first: which error each belongs to, when it was received, its exception class and message, severity, whether it was unhandled, and its release stage. Returns a bare array. Narrow it with since, before and the Event Field filters. Set full_reports for the complete event payload including metadata, request, device and user.

Bugsnag: List Project Pivot Values

Break a whole project down along one dimension: every value of the chosen Event Field with its event count, share of the total, and first and last seen times. The distribution question, e.g. how events split across releases or devices. Returns a bare array.

Bugsnag: List Project Pivots

Summarize a whole project across every dimension: for each Event Field, how many unique values it has and which top values account for the events. The project-wide view of where errors are concentrated. Returns a bare array.

Bugsnag: List Projects

List an organization's projects with their IDs, slugs, framework types, notifier API keys, release stages and open error counts. Returns a bare array. The project ID from here feeds every error, event, pivot, trend and release node.

Bugsnag: List Releases

List a project's releases, newest first: app version, build label, who or what built it, source control revision, how many errors it introduced and saw, and its session counts. The release-health read behind questions like whether the last deploy made things worse. Returns a bare array.

Action nodes 12

Bugsnag: Bulk Update Errors

Apply one workflow operation to many errors at once: the same set as Bugsnag: Update Error (fix, open, ignore, snooze, assign, override_severity, create_issue, link_issue, unlink_issue, discard). The response echoes the operation and the resulting state of each error ID. Reversible with the inverse operation over the same IDs. Needs project owner or organization administrator permissions.

Bugsnag: Comment on Error

Post a comment on an error as the token's user. The usual automation move is writing back what a workflow did: which deploy is suspected, which ticket was opened, who was paged. Reversible in the Bugsnag dashboard, which is where comment deletion lives.

Bugsnag: Create Project

Create a project in an organization and get back its notifier API key, which is what an app's Bugsnag library is configured with. Reversible with Bugsnag: Delete Project while the project is still empty.

Bugsnag: Delete All Errors in Project

DESTRUCTIVE: delete ALL error and event data in a project. The vendor documents this as an action that cannot be reversed. It takes no filters, so there is no partial version of it. Needs project owner or organization administrator permissions.

Bugsnag: Delete Error

DESTRUCTIVE: delete one error and every event grouped into it. Its comments, assignment, linked issues and history go with it. If the same crash happens again it comes back as a NEW error with status open and no history. Needs project owner or organization administrator permissions.

Bugsnag: Delete Event

DESTRUCTIVE: delete one event occurrence, including its stack trace, metadata and any user data captured with it. There is no undelete. The parent error survives with one fewer occurrence.

Bugsnag: Delete Project

DESTRUCTIVE: delete a project along with every error and event in it. The notifier API key stops working, so any app still reporting to it goes dark. There is no undelete. Requires organization administrator permissions, and projects maintained by a third-party add-on cannot be deleted at all.

Bugsnag: Invite Collaborator

Invite someone to the organization by email. Bugsnag SENDS AN INVITATION EMAIL to the address, so point automated runs at a test address you own. Grant project access at the same time with project_ids or project_roles. Reversible with Bugsnag: Delete Collaborator.

Bugsnag: Remove Collaborator

DESTRUCTIVE: remove a person from the organization. They lose access to every project in it, and their team memberships and project roles are gone; re-adding them means a fresh invitation and rebuilding those grants. Errors they were assigned to become unassigned.

Bugsnag: Update Collaborator Permissions

Change which projects a collaborator can reach, their role on each, and whether they administer the organization. project_ids REPLACES the access list, so read the current state with Bugsnag: List Collaborators first and send the full set. Reversible by writing the previous values back.

Bugsnag: Update Error

Change one error's workflow state. The operation field picks what happens: fix, open, ignore, snooze (with reopen_rules), assign (to assigned_collaborator_id, or blank to unassign), override_severity (with severity), create_issue, link_issue (with issue_url), unlink_issue, discard or undiscard. Every state has an inverse operation, so this is reversible. The operation named delete is NOT: use Bugsnag: Delete Error, which says so in its name. Needs project owner or organization administrator permissions.

Bugsnag: Update Project

Edit a project's name, grouping rules, discard rules and collaborator list. Only the fields you fill are sent. Array fields REPLACE their current contents, and collaborator_ids removes anyone not listed, so read the current values with Bugsnag: Get Project first. Changing grouping rules does not regroup existing errors. Reversible by writing the previous values back.

Connecting Bugsnag takes one credential

Bugsnag authenticates Data Access API requests with a personal auth token, sent as an Authorization header carrying Bugsnag's literal token prefix rather than Bearer. The token is generated in Bugsnag under My account, then Personal auth tokens, and it acts as you: it can see and do what your Bugsnag user can see and do, across every organization your account belongs to. One extra field matters at setup. Bugsnag documents a different base server URL per dashboard: app.bugsnag.com maps to api.bugsnag.com, app.bugsnag.smartbear.com maps to api.bugsnag.smartbear.com, and an on-premise installation maps to api.your-domain. Point the credential at the host that matches the dashboard you sign in at.

1

Sign in to Bugsnag, open My account, then Personal auth tokens, and create a token. Copy it when it is shown.

2

In Mattering Automations, open Connectors, click Add Connector, choose Bugsnag, and create a credential. Paste the token into Personal Auth Token.

3

Set API Host only if your organization signs in somewhere other than app.bugsnag.com: use api.bugsnag.smartbear.com for app.bugsnag.smartbear.com, or api.your-domain for an on-premise installation. Leave it blank for app.bugsnag.com organizations; a dashboard domain is accepted and mapped to the matching API host.

4

Click Test Connection. Mattering lists the organizations the token can see, which is the cheapest authenticated read in the API and needs no project context. A rejection with the host named in the message usually means the credential points at a different host than the dashboard the token was created on, so check that value before replacing the token.

5

All 35 Bugsnag nodes appear in the canvas palette. Run Bugsnag: List Organizations first for the organization ID, then Bugsnag: List Projects for the project ID that every error, event, pivot, trend and release node asks for.

Common uses for the Bugsnag MCP

From an alert to the actual stack trace in one hop

The slow part of triage is getting from a notification to the evidence. Bugsnag: Get Latest Event on Error returns the most recent occurrence of an error as a full report, stack trace, request, device and user context included, in a single call with no list to page through first. When one occurrence is not enough, Bugsnag: List Error Events walks the occurrences newest first and full_reports pulls complete payloads, and Bugsnag: Get Event fetches a specific one with its breadcrumbs, threads, feature flags and custom metadata. That is the raw material a workflow can paste into a task or hand to an agent to summarize.

Triage that writes back instead of just reading

Bugsnag: Update Error is the whole workflow verb set behind one operation field: fix, open, ignore, snooze with reopen rules, assign to a collaborator or blank to unassign, override severity, create or link or unlink an issue, discard and undiscard. Every state has an inverse, so an automation that snoozes noise overnight can unsnooze it in the morning. Bugsnag: Bulk Update Errors applies one operation across many error IDs for the weekly sweep, and Bugsnag: Comment on Error records what the workflow did, which deploy is suspected, which ticket was opened, who was paged, so the next person reads the reasoning on the error itself.

Release health that a schedule can watch

Bugsnag: Get Project Stability Trend returns one point per UTC day for the last 30 days on the project's primary release stage: total sessions, sessions that saw an unhandled error, users seen, and users who hit one. Bugsnag: List Releases adds the per-release view, how many errors each release introduced and saw against its session counts, and Bugsnag: Get Release drills into one. A scheduled sweep can write the series into a Mattering database and open a task when stability drops after a deploy, which is the question people actually ask a crash reporter.

Charts from trends, without inventing the buckets

Bugsnag: Get Project Trend and Bugsnag: Get Error Trend return time-bucketed event counts over a window you set with since and before, both of which accept Bugsnag's relative form as well as ISO 8601. Choose either buckets_count, which splits the window into that many equal buckets, or resolution, which fixes the bucket width. Each row carries its own from and to timestamps and an events_count, so the output is directly chartable and additive across adjacent buckets, and the same filters that narrow an error list narrow the trend, which is how you get unhandled errors in production per hour rather than everything per hour.

Where an error actually lands, by any dimension

Pivots answer the question a stack trace cannot. Bugsnag: List Error Pivots summarizes one error across every dimension Bugsnag tracks, showing for each Event Field how many unique values it has and which values carry the occurrences, and Bugsnag: List Error Pivot Values returns the full ranked list for one chosen dimension with counts, shares, and first and last seen times. Bugsnag: List Project Pivots and Bugsnag: List Project Pivot Values do the same project-wide. Run Bugsnag: List Event Fields first: it lists the display IDs a project supports, which are also the keys the filters on the error, event, pivot and trend nodes accept.

Project and access administration as workflow steps

Bugsnag: Create Project stands up a project and returns the notifier API key an app's Bugsnag library is configured with, so provisioning a new service can include its crash reporting. Bugsnag: Update Project edits grouping and discard rules, with the caveat the node states, changing grouping rules does not regroup existing errors. On the people side, Bugsnag: Invite Collaborator grants access and sends an invitation email, so point automated runs at an address you own, Bugsnag: Update Collaborator Permissions changes project access and roles, and Bugsnag: List Project Collaborators is the right list to pick an assignee from. Bugsnag does not reject an assignment to someone without access to the project: it returns success and leaves the error unassigned, and it does the same for a collaborator who has not accepted their invitation, so pick the target from the project list and read the assignment back rather than trusting the status code.

Example: A new error shows up in production: gather the evidence, assign it, open the task

Built on the Automations canvas in a few minutes, or described in plain language to Matter and built for you.

1

Trigger: a schedule runs every fifteen minutes, or an inbound webhook fires from your alerting.

2

Bugsnag: List Errors reads the project with since set to a short relative window and a filter of error.status eq open, so only errors that are still live come back.

3

Bugsnag: Get Latest Event on Error pulls the freshest full report for the top error: stack trace, request, device and user context in one call.

4

Bugsnag: List Error Pivot Values breaks the same error down by app version, which usually names the release that introduced it.

5

Bugsnag: Get Error Trend returns the bucketed occurrence counts for the window, so the task carries a shape rather than a single number.

6

The workflow opens a task in Mattering Productivity with the stack trace, the version breakdown and the trend attached, and assigns it to the team that owns the service.

7

Bugsnag: Update Error assigns the error to a collaborator from Bugsnag: List Project Collaborators, and Bugsnag: Comment on Error writes the task link back onto the error so both sides point at each other.

Bugsnag MCP questions, answered

What Bugsnag operations are supported?

35 nodes on the Bugsnag Data Access API v2: 23 query nodes and 12 action nodes. Errors (list with the filters DSL, get, update through the full workflow verb set, bulk update, delete, delete all in a project), error comments (list and create), events (project and per-error listers, get one, get the latest on an error, delete), pivots and pivot values for both an error and a project, the project's Event Field vocabulary, windowed error and project trends plus the 30 day session stability trend, releases (list and get), projects (list, get, create, update, delete), organizations (list and get), and collaborators (organization and project listers, invite, update permissions, remove). Deliberately not in this wave: performance monitoring (spans, traces, trace fields), teams, SCIM user and group provisioning, saved searches, feature flags, configured integrations, GDPR and CCPA data requests, and account administration such as API key and auth token regeneration.

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.

Is there an official Bugsnag MCP server?

Yes. SmartBear publishes a SmartBear MCP Server with a documented BugSnag integration, and if your only goal is asking an assistant about your errors in a chat client, that is a reasonable route. This connector solves a different problem. The same Bugsnag operations sit on a visual automation canvas next to every other connector, so a Bugsnag step can run on a schedule or behind a webhook, chain into a Productivity task, a database write, or a human approval, and be called by an assistant over MCP using the same credential and the same node inventory. It is the difference between a chat client that can read Bugsnag and a workspace where Bugsnag is one step in a workflow that keeps running when nobody is typing.

What credential does it use, and why is there a host field?

A Bugsnag personal auth token, generated under My account, then Personal auth tokens, and sent in an Authorization header with Bugsnag's literal token prefix rather than Bearer. The connector also pins the API to version 2 with Bugsnag's documented version header. The host field exists because Bugsnag documents a different base server URL per dashboard: api.bugsnag.com for organizations that sign in at app.bugsnag.com, api.bugsnag.smartbear.com for organizations on app.bugsnag.smartbear.com, and api.your-domain for an on-premise installation. Point the credential at the host that matches the dashboard you sign in at, and if a request is rejected check that value before you replace the token. Leave the field blank for app.bugsnag.com. The token acts as your Bugsnag user, so the connector can do what you can do and nothing more.

How does pagination work on the list nodes?

This is a real limitation and every lister's description says so rather than hiding it. Bugsnag paginates with the Link response header and reports the total in the X-Total-Count header, not with a cursor or a page number in the response body. The connector's node output carries the parsed body, so most listers cannot hand the next step a next-page URL. Releases are the exception, because that endpoint takes an explicit offset a workflow can increment itself. Elsewhere the practical advice is to raise per_page rather than page. What Bugsnag documents: 30 results per page by default, a custom page size of up to 100 on resources that allow it, and a lower maximum stated in that endpoint's own reference where one applies. Releases stay the documented exception at 1 to 10 per page with an offset, and their default is 5. When you need a narrower result rather than a longer one, use since, before and the Event Field filters, which is usually the better instrument anyway.

What are Bugsnag's API rate limits?

Bugsnag does not publish a numeric request limit for the Data Access API, so this page will not invent one. What its documentation does describe is the mechanism: a per-minute window, rate limit headers on successful responses telling you the limit and how much of it is left, and an HTTP 429 with a Retry-After value in seconds when the window is exhausted. The nodes report an error on their output rather than silently retrying, so a workflow can branch, wait, and continue rather than failing the run.

Can an AI agent safely resolve, ignore, or delete errors?

Resolving and ignoring, yes, and that is the designed-for job. Bugsnag routes every workflow change through Bugsnag: Update Error, where each operation has an inverse: fix and open, ignore and open, snooze and open, discard and undiscard, assign with a blank collaborator to unassign. An agent that gets it wrong can be told to put it back. Deleting is a different class. Bugsnag: Delete Error removes one error and the events grouped into it, and if the same crash happens again it comes back as a new error with status Open rather than reopening the old one. Bugsnag: Delete Event removes a single occurrence. Bugsnag: Delete All Errors in Project takes no filters and is documented by Bugsnag as an action that cannot be reversed. All five destructive nodes are flagged as such, and the honest pattern is to keep them out of automated paths or put them behind a human approval step.

Do I need a particular Bugsnag plan for the API?

Personal auth tokens are generated under My account, then Personal auth tokens, and this connector uses the same public Data Access API that Bugsnag documents for exactly that purpose. Being able to sign in is not on its own enough: Bugsnag documents API error codes for a lapsed subscription, for an expired trial, and for a user with no available seat, and its pricing comparison lists the data export API as a plan feature with volume tiers. This page does not state which Bugsnag plans include or restrict API access, because plan terms change and the vendor's pricing page is the only place worth reading them. One signal is worth knowing at build time: Bugsnag documents specific API error codes for those restrictions, so if your account's plan or your seat limits API access you will get one of those rather than a connector fault, and the node reports it on its output.

35 nodes waiting on the canvas

Put Bugsnag 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. Bugsnag is a product of its respective owner; Mattering is an independent platform.

Start Free Explore our Products