vignesh07/claude-on-call
AI-powered on-call agent that responds to production alerts with analysis and fixes
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| anthropic_api_key | Anthropic API key for Claude | yes | — |
| alert_payload | JSON alert payload from webhook | yes | — |
| alert_source | Source of alert (pagerduty, datadog, cloudwatch, sentry, opsgenie, prometheus, generic, auto) | no | auto |
| mode | Operation mode: pr = attempt fix, analyze = comment only, auto = decide based on context, review = respond to PR comments | no | auto |
| create_issue | Create GitHub issue to track the alert | no | true |
| pagerduty_api_key | PagerDuty API key for updating incidents (optional) | no | — |
| confidence_threshold | Minimum confidence to create PR (high, medium, low) | no | medium |
| timeout_minutes | Maximum time for Claude to work (minutes) | no | 10 |
| max_files_changed | Maximum number of files Claude can modify in a fix | no | 10 |
| draft_pr | Create pull requests as drafts | no | true |
| test_command | Command to run tests (e.g., npm test, pytest). If set, Claude will run tests and fix failures. | no | — |
| pr_number | PR number to respond to (for review mode) | no | — |
| comment_body | Comment body to respond to (for review mode) | no | — |
| comment_id | Comment ID for threading responses (for review mode) | no | — |
Outputs
| name | description |
|---|---|
| action_taken | What the agent did: pr_created, analysis_only, duplicate, error |
| pr_number | Pull request number if created |
| issue_number | GitHub issue number for tracking |
| analysis | Agent analysis summary |
| confidence | Agent confidence in the fix: high, medium, low |
| duplicate_of | Issue number if this alert was detected as a duplicate |