Investigate Sentry incidents with Google Gemini, Postgres, GitHub and Slack

## Autonomous Production Incident Investigator ### How it works This workflow receives Sentry issue alerts, normalizes the alert data, and checks PostgreSQL to avoid reprocessing duplicate incidents. For new incidents, it enriches the alert with Sentry issue details, recent GitHub commits, and historical incident records before assembling an evidence bundle for an AI agent. The AI investigation is validated, scored, saved back to PostgreSQL, summarized, and sent to Slack for responders. @youtube ### Setup steps - Configure the Sentry webhook to call the workflow’s Webhook URL and ensure the payload includes the issue identifiers expected by the normalization code. - Set up Sentry credentials for fetching issue details. - Configure the GitHub HTTP request with the target repository, API authentication if required, and any commit filtering parameters. - Configure PostgreSQL credentials and ensure the incident/history tables used by the duplicate check, historical lookup, and upsert queries exist. - Configure Google Gemini credentials for the AI Agent chat model. - Configure Slack credentials and select the destination channel for incident alerts. ### Customization Adjust the normalization code, duplicate-detection query, GitHub repository URL, AI investigation prompt, scoring thresholds, and Slack summary format to match your incident response process.