Bugsnag MCP Server

Bugsnag

Error monitoring platform for tracking, diagnosing, and resolving application errors. Explore the org → project → error/event hierarchy to investigate stability issues.

Entity Hierarchy

Organization Level

• Organizations — top-level container. Resolve organization_id first — it anchors everything below.
• Collaborators — team members with access to the org. Scoped by organization_id.
• Teams — groups within the org for permission management. Scoped by organization_id.
• Supported Integrations — available third-party integrations (Slack, Jira, etc.). Global list, no ID needed.

Project Level (requires organization, then project context)

• Projects — applications being monitored. Enumerate under organization_id to discover project_id.
• Errors — grouped error classes (e.g., "NullPointerException in UserService.getUser"). Each error aggregates many individual events. Sort by last_seen, events, users, or created_at. Filter by status (open/fixed/snoozed/ignored). Open one error class by id when you need its aggregate view. Scoped by project_id.
• Events — individual error occurrences with full stack traces, device info, and metadata. Use full_reports: true for complete payloads including stack traces. Scoped by project_id. Events are instances of errors.
• Event Fields — available metadata fields for filtering events (app version, OS, etc.). Scoped by project_id.
• Pivots — error grouping breakdowns (by release stage, browser, device, etc.). Scoped by project_id.
• Releases — deployed versions of your app, with error counts and stability scores. Scoped by project_id.
• Trace Fields — available trace/span fields for distributed tracing. Scoped by project_id.

Saved Search Level (requires project, then saved-search id)

• Saved Searches — stored error filters/queries within a project. Scoped by project_id.
• Saved Search Usage Summary — how frequently a saved search is used. Requires search_id from saved searches.

Workflows

• Error investigation: Organizations → projects for the org → error groups sorted by last_seen → one error id for the class → event feed with full_reports: true when you need stacks.
• Release stability: Projects → release history to compare crash density across versions.
• Team audit: Organizations → collaborators and teams for access review.
• Search analysis: Saved searches → usage summaries to see which stored filters fire most.

Gotchas

• Three-level ID chain: organization_id → project_id → search_id. Each level requires the parent ID from the level above.
• Errors vs Events: errors are grouped classes; events are individual occurrences. Scan error groups for overview, drill into events when you need stacks and occurrence detail.
• Pagination: Long lists use offset + per_page style paging.