Cursor Automations + Herald

On-call

Trigger: PagerDuty incident, cron health check, or manual webhook

When on-call escalation is warranted:
1. Assess severity from context (production down = critical, degraded = warning).
2. Call `pagerduty_trigger_incident` with summary, severity, and details.
3. Notify the team channel with `discord_send_notification` including incident link and next steps.
Do not page for low-severity or unconfirmed issues.

Web3 watch

Trigger: Herald inbox webhook (Alchemy / Tenderly / Forta) or cron monitor

When suspicious on-chain activity appears in Herald inbox:
1. Call `list_recent_events` then `get_event` for the relevant monitor event.
2. Call `discord_send_web3_alert` or `slack_send_web3_alert` with severity, chain, tx_hash, block, title, summary, explorer_url.
Use critical severity for exploit patterns or large unexpected outflows.

Solana watch

Trigger: Helius webhook (via Herald inbox) or scheduled Solana monitor

When a Solana monitor fires:
1. Read the event from Herald inbox with `list_recent_events` / `get_event`.
2. Call `discord_send_solana_alert` with chain, signature, slot, title, summary, explorer_url, and severity.

Stripe ops

Trigger: Stripe webhook (via Herald inbox) or scheduled billing check

When a Stripe payment or billing event indicates failure:
1. Read `list_recent_events` for the latest stripe-sourced event if needed.
2. Call `discord_send_stripe_alert` or `slack_send_stripe_alert` with title, summary, event_type, customer_or_account, amount_or_plan, severity, and dashboard context.
Use severity high for repeated failures; medium for single recoverable failures.

Ship guard

Trigger: GitHub — CI completed (failed) or Pull request opened

Point GitHub Actions or Vertel webhooks at Herald Inbox for richer context.

When this automation runs:
1. If CI failed or deploy status is failure, call `discord_send_ci_alert` with provider, repo_or_project, workflow_or_env, status, run_url, severity, title, and summary.
2. If severity is critical, also call `pagerduty_trigger_incident` with summary and details.
3. If unsure, call `list_recent_events` first to read the latest GitHub Actions or Vercel inbox event, then alert.
Do nothing if CI passed and there is no actionable failure.

Status room

Trigger: Sentry — Issue created (via Herald webhook) or cron every 15 minutes

Connect Sentry webhooks to Herald Inbox under Event sources.

When a Sentry-style error spike or new issue is detected:
1. Call `list_recent_events` and `get_event` to read the latest Sentry inbox event if event_id is not provided.
2. Call `slack_send_sentry_alert` with title, summary, project, level, issue_url, severity, and severity_color.
3. For customer-facing outages, call `statuspage_create_incident` with name, status, and body per schema.
Escalate only when user impact is confirmed or severity is high/critical.

Support desk

Trigger: Zendesk / Intercom webhook or manual triage request

When a support escalation is needed:
1. Create a ticket with `zendesk_create_ticket` or `intercom_create_conversation` using required fields from the tool schema.
2. Include customer context, severity, and recommended next action in the body.

Calendar agent

Trigger: Cal.com booking webhook or scheduled reminder

When a Cal.com event requires action:
1. Read inbox events if the booking came via Herald webhook.
2. Call `calcom_create_booking` when scheduling is requested.
3. Notify with `slack_send_message` confirming the booking or handoff.

Dev ship

Trigger: GitHub — Pull request merged or Issue created

After a PR is merged or when triage is requested:
1. If follow-up work is needed, call `github_create_issue` with owner, repo, title, and body summarizing what shipped and what remains.
2. Notify the team with `slack_send_message` including the issue link and one-line summary.
Skip if the PR is docs-only with no follow-ups.