| name | rollbar |
| description | Query Rollbar error tracking to investigate errors, list recent issues, resolve items, and get context for bug fixes. Use when the user asks about production errors, wants to investigate an error, asks "what errors happened overnight", wants to mark issues as resolved, or needs context to fix a bug from Rollbar. |
| user-invocable | false |
| allowed-tools | Bash(rollbar *) |
Rollbar Error Investigation
Use the rollbar CLI to query and manage Rollbar errors. This tool is optimized for investigating production errors, providing context for bug fixes, and resolving issues.
Prerequisites
Ensure ROLLBAR_ACCESS_TOKEN is set in the environment, or a .rollbar.yaml config file exists in the project.
Common Workflows
Check what errors happened recently
rollbar items --since "8 hours ago" --level error,critical
rollbar items --since "12 hours ago" --level error,critical --env production
rollbar items --status active --limit 10
Investigate a specific error
rollbar item 123
rollbar context 123
rollbar context 123 --occurrences 5
The context output includes (when available):
- Exception class, message, and full stack trace
- Request URL, method, and the user's browser (User-Agent header)
- Logged-in user's email address and ID
- Server host, branch, and code version
Search for specific errors
rollbar items --query "TypeError" --level error
rollbar items --env production --level error --since "24h"
rollbar items --sort occurrences --limit 10
View error occurrences
rollbar occurrences --item 123 --limit 5
rollbar occurrence 453568801204
Resolve items
Mark items as resolved after fixing the underlying issue (requires write token):
rollbar resolve 123
rollbar resolve 93 95 97
rollbar resolve --uuid 8675309
Output Formats
--output table (default): Human-readable tables
--output json: Full JSON for parsing
--output compact or --ai: Token-efficient format for AI context
--output markdown: Structured markdown
Key Flags
| Flag | Description |
|---|
--level | Filter by level: debug, info, warning, error, critical |
--status | Filter by status: active, resolved, muted, any |
--env | Filter by environment (e.g., production, staging) |
--since | Time filter: "8 hours ago", "24h", "7 days" |
--query | Text search in item titles |
--sort | Sort by: recent, occurrences, first-seen, level |
--limit | Limit number of results |
Typical Investigation Flow
- List recent errors:
rollbar items --since "8 hours ago" --level error,critical
- Identify the issue: Look at error titles and occurrence counts
- Get full context:
rollbar context <item-number> for the specific error
- Analyze stack trace: The context output includes file paths and line numbers
- Check patterns: Look at multiple occurrences to understand the trigger
- Fix and resolve: After deploying a fix, mark the item as resolved:
rollbar resolve <item-number>
Example: Overnight Error Triage
When asked "what errors happened overnight?" or similar:
rollbar items --since "12 hours ago" --level error,critical --env production --ai
rollbar context <item-number>
The --ai flag provides compact output that's token-efficient while preserving all essential information.
Example: Resolve Items After Fix
When asked "mark items 93 and 95 as resolved" or similar:
rollbar resolve 93 95
Note: Resolving items requires a project access token with "write" scope. If you get a permission error, the token only has read access.