Every customer gets a daily health score with actionable recommendations for at-risk accounts

Pull subscription status and payment history from Stripe. Pull usage metrics (messages sent, agents run, jobs executed) from the database. Pull error counts and latency from monitoring tools. For each paying customer, compile a snapshot row.

For each account, classify based on:
- **Usage trend**: increasing, flat, or declining over the last 7 and 30 days
- **Error health**: error rate relative to usage volume, any spikes
- **Payment health**: current, past-due, or failed; any recent plan changes
- **Engagement**: recent logins, new agent creations, new integrations connected

Assign a tier: **Hot** (growing, engaged, clean), **Warm** (steady, no red flags), **Cold** (declining, errors, payment issues).

Sort accounts by action urgency:
- Cold accounts with payment issues at the top
- Cold accounts with declining usage next
- Hot accounts with expansion signals (approaching limits, exploring new features)
- Warm accounts last

For each account in the top 20, draft a one-line recommendation (e.g. 'Send a check-in email', 'Investigate error spike from yesterday', 'Offer a plan upgrade').

Post the ranked briefing to #fde. Include a summary counts (X hot, Y warm, Z cold), the top 10 accounts by urgency with recommendations, and a section on expansion opportunities.