sentry-automation（哨兵自动化）

Sentry Automation via Rube MCP

Automate Sentry error tracking and monitoring operations through Composio's Sentry toolkit via Rube MCP.

Prerequisites

• Rube MCP must be connected (RUBE_SEARCH_TOOLS available)
• Active Sentry connection via RUBE_MANAGE_CONNECTIONS with toolkit sentry
• Always call RUBE_SEARCH_TOOLS first to get current tool schemas

Setup

Get Rube MCP: Add https://rube.app/mcp as an MCP server in your client configuration. No API keys needed — just add the endpoint and it works.

1. Verify Rube MCP is available by confirming RUBE_SEARCH_TOOLS responds
2. Call RUBE_MANAGE_CONNECTIONS with toolkit sentry
3. If connection is not ACTIVE, follow the returned auth link to complete Sentry OAuth
4. Confirm connection status shows ACTIVE before running any workflows

Core Workflows

1. Investigate Issues

When to use: User wants to find, inspect, or triage error issues

Tool sequence:

1. SENTRY_LIST_AN_ORGANIZATIONS_ISSUES - List issues across the organization [Required]
2. SENTRY_GET_ORGANIZATION_ISSUE_DETAILS - Get detailed info on a specific issue [Optional]
3. SENTRY_LIST_AN_ISSUES_EVENTS - View individual error events for an issue [Optional]
4. SENTRY_RETRIEVE_AN_ISSUE_EVENT - Get full event details with stack trace [Optional]
5. SENTRY_RETRIEVE_ISSUE_TAG_DETAILS - Inspect tag distribution for an issue [Optional]

Key parameters:

• organization_id_or_slug: Organization identifier
• issue_id: Numeric issue ID
• query: Search query (e.g., is:unresolved, assigned:me, browser:Chrome)
• sort: Sort order (date, new, freq, priority)
• statsPeriod: Time window for stats (24h, 14d, etc.)

Pitfalls:

• organization_id_or_slug is the org slug (e.g., 'my-org'), not the display name
• Issue IDs are numeric; do not confuse with event IDs which are UUIDs
• Query syntax uses Sentry's search format: is:unresolved, assigned:me, !has:release
• Events within an issue can have different stack traces; inspect individual events for details

2. Manage Project Issues

When to use: User wants to view issues scoped to a specific project

Tool sequence:

1. SENTRY_RETRIEVE_ORGANIZATION_PROJECTS - List projects to find project slug [Prerequisite]
2. SENTRY_RETRIEVE_PROJECT_ISSUES_LIST - List issues for a specific project [Required]
3. SENTRY_RETRIEVE_ISSUE_EVENTS_BY_ID - Get events for a specific issue [Optional]

Key parameters:

• organization_id_or_slug: Organization identifier
• project_id_or_slug: Project identifier
• query: Search filter string
• statsPeriod: Stats time window

Pitfalls:

• Project slugs are different from project display names
• Always resolve project names to slugs via RETRIEVE_ORGANIZATION_PROJECTS first
• Project-scoped issue lists may have different pagination than org-scoped lists

3. Configure Alert Rules

When to use: User wants to create or manage alert rules for a project

Tool sequence:

1. SENTRY_RETRIEVE_ORGANIZATION_PROJECTS - Find project for the alert [Prerequisite]
2. SENTRY_RETRIEVE_PROJECT_RULES_BY_ORG_AND_PROJECT_ID - List existing rules [Optional]
3. SENTRY_CREATE_PROJECT_RULE_FOR_ALERTS - Create a new alert rule [Required]
4. SENTRY_CREATE_ORGANIZATION_ALERT_RULE - Create org-level metric alert [Alternative]
5. SENTRY_UPDATE_ORGANIZATION_ALERT_RULES - Update existing alert rules [Optional]
6. SENTRY_RETRIEVE_ALERT_RULE_DETAILS - Inspect specific alert rule [Optional]
7. SENTRY_GET_PROJECT_RULE_DETAILS - Get project-level rule details [Optional]

Key parameters:

• name: Alert rule name
• conditions: Array of trigger conditions
• actions: Array of actions to perform when triggered
• filters: Array of event filters
• frequency: How often to trigger (in minutes)
• actionMatch: 'all', 'any', or 'none' for condition matching

Pitfalls:

• Project-level rules (CREATE_PROJECT_RULE) and org-level metric alerts (CREATE_ORGANIZATION_ALERT_RULE) are different
• Conditions, actions, and filters use specific JSON schemas; check Sentry docs for valid types
• frequency is in minutes; setting too low causes alert fatigue
• actionMatch defaults may vary; explicitly set to avoid unexpected behavior

4. Manage Releases

When to use: User wants to create, track, or manage release versions

Tool sequence:

1. SENTRY_LIST_ORGANIZATION_RELEASES - List existing releases [Optional]
2. SENTRY_CREATE_RELEASE_FOR_ORGANIZATION - Create a new release [Required]
3. SENTRY_UPDATE_RELEASE_DETAILS_FOR_ORGANIZATION - Update release metadata [Optional]
4. SENTRY_CREATE_RELEASE_DEPLOY_FOR_ORG - Record a deployment for a release [Optional]
5. SENTRY_UPLOAD_RELEASE_FILE_TO_ORGANIZATION - Upload source maps or files [Optional]

Key parameters:

• version: Release version string (e.g., '1.0.0', commit SHA)
• projects: Array of project slugs this release belongs to
• dateReleased: Release timestamp (ISO 8601)
• environment: Deployment environment name (e.g., 'production', 'staging')

Pitfalls:

• Release versions must be unique within an organization
• Releases can span multiple projects; use the projects array
• Deploying a release is separate from creating it; use CREATE_RELEASE_DEPLOY
• Source map uploads require the release to exist first

5. Monitor Organization and Teams

When to use: User wants to view org structure, teams, or member lists

Tool sequence:

1. SENTRY_GET_ORGANIZATION_DETAILS or SENTRY_GET_ORGANIZATION_BY_ID_OR_SLUG - Get org info [Required]
2. SENTRY_LIST_TEAMS_IN_ORGANIZATION - List all teams [Optional]
3. SENTRY_LIST_ORGANIZATION_MEMBERS - List org members [Optional]
4. SENTRY_GET_PROJECT_LIST - List all accessible projects [Optional]

Key parameters:

• organization_id_or_slug: Organization identifier
• cursor: Pagination cursor for large result sets

Pitfalls:

• Organization slugs are URL-safe identifiers, not display names
• Member lists may be paginated; follow cursor pagination
• Team and member visibility depends on the authenticated user's permissions

6. Manage Monitors (Cron Monitoring)

When to use: User wants to update cron job monitoring configuration

Tool sequence:

1. SENTRY_UPDATE_A_MONITOR - Update monitor configuration [Required]

Key parameters:

• organization_id_or_slug: Organization identifier
• monitor_id_or_slug: Monitor identifier
• name: Monitor display name
• schedule: Cron schedule expression or interval
• checkin_margin: Grace period in minutes for late check-ins
• max_runtime: Maximum expected runtime in minutes

Pitfalls:

• Monitor slugs are auto-generated from the name; use slug for API calls
• Schedule changes take effect immediately
• Missing check-ins trigger alerts after the margin period

Common Patterns

ID Resolution

Organization name -> Slug:

1. Call SENTRY_GET_ORGANIZATION_DETAILS
2. Extract slug field from response

Project name -> Slug:

1. Call SENTRY_RETRIEVE_ORGANIZATION_PROJECTS
2. Find project by name, extract slug

Pagination

• Sentry uses cursor-based pagination with Link headers
• Check response for cursor values
• Pass cursor in next request until no more pages

Search Query Syntax

• is:unresolved - Unresolved issues
• is:resolved - Resolved issues
• assigned:me - Assigned to current user
• assigned:team-slug - Assigned to a team
• !has:release - Issues without a release
• first-release:1.0.0 - Issues first seen in release
• times-seen:>100 - Seen more than 100 times
• browser:Chrome - Filter by browser tag

Known Pitfalls

ID Formats:

• Organization: use slug (e.g., 'my-org'), not display name
• Project: use slug (e.g., 'my-project'), not display name
• Issue IDs: numeric integers
• Event IDs: UUIDs (32-char hex strings)

Permissions:

• API token scopes must match the operations being performed
• Organization-level operations require org-level permissions
• Project-level operations require project access

Rate Limits:

• Sentry enforces per-organization rate limits
• Implement backoff on 429 responses
• Bulk operations should be staggered

Quick Reference