Sentinel-aaS onboarding — Day 0 to Day 30.

You complete checkout

From the Sentinel-aaS page you click "Start Sentinel Scale — $5,999/mo" (or Pro / Enterprise). Stripe handles payment. garnetgrid-fulfillment creates your envelope at garnet-tokens/sentinel/<slug>.json and sends a welcome email.

The welcome email asks one question that's specific to this lane: Which Discord workspace? If you have one, share an invite link to a fresh admin server (we'll provision the bot there). If you don't, we'll create a fresh workspace as part of onboarding (Pro tier includes one).

Integration list, channel layout, escalation roles

The intake call is where we figure out the bus shape:

1. Integration list — which webhook sources go through the bus. Typical day-1 set: Stripe (payments), GitHub (PRs + CI), Calendly (bookings), plus 1–2 of (Datadog | Sentry | Grafana | PagerDuty).
2. Channel layout — how the workspace channels map to lane-aware routing. Common pattern: #leads, #revenue, #engineering, #alerts-p0, #alerts-p1, #deploys.
3. Escalation roles — Discord role for "operator-on-call" and "leadership". These get @-mentioned by the bot on P0 incidents. We define 2–4 routing rules during the call ("if Stripe dispute → ping ops-team", "if Datadog severity=critical and source=auth-svc → ping operator-on-call").

Sentinel bot deploys into your Discord workspace

On the operator side: bot provisioning takes ~30 minutes.

• Sentinel bot installed in your workspace with the necessary permissions (manage channels, send messages with embeds, use slash-commands, mention roles).
• Cloudflare Worker per integration deployed in your Cloudflare account. Worker names: sentinel-stripe, sentinel-github, sentinel-calendly, etc.
• R2 audit-log bucket created at garnet-snapshots/sentinel/<slug>/events/. Every webhook payload + bot action lands here, retained indefinitely.
• Slash-command suite registered: /sentinel-status, /replay, /quiet, /lead-status, /route-test.
• Heartbeat probes scheduled (Cron Trigger every 5 minutes per integration).

Real webhooks land in lane-aware channels

Within the first 24 hours of bot activation, real webhooks start landing in your channels. Sentinel routes per the day-1 rules, fans embed-rich messages to the right channel, and writes the full audit log to R2.

Engineer reviews routing metrics, prunes false-positives

First week's traffic surfaces what's actually noisy. The engineer reviews:

• Lead → operator latency histogram (target: <60s for hot, <5min for warm)
• Slash-command usage (which commands are unused, which need response-shape tuning)
• Alert sources firing >10×/day (rate-limit or filter)
• Co-firing alert signatures (cluster into single embed)

Tuning ships as Worker code commits — the bot doesn't get config-edited, it gets redeployed. Routing rules are versioned in your Cloudflare account.

Sentinel-monthly Workflow renders + emails the PDF

On the 1st of next month, garnet-sentinel-monthly Workflow fires. Aggregates the month's events, computes integration uptime, surfaces top alerts + median resolution times, slash-command usage histogram, then renders the PDF.

The PDF covers:

• Event volume — total webhooks routed, breakdown per integration
• Integration health — per-integration uptime % + downtime windows + notes
• Routing metrics — rules fired, leads routed, median lead-to-operator latency
• Top alerts — most-frequent alert signatures + median resolution time
• Slash-command usage — top-N commands + invocation count
• Uptime — overall bot uptime % + incidents (with cause + fix)

Steady-state cadence from here: daily traffic flowing, weekly routing review + tuning, monthly executive PDF.