| name | rhombus-cross-client-reporting |
| description | Produce fleet-wide health and activity reports across every managed Rhombus client organization — weekly health summaries, monthly executive reports, incident rollups, license utilization, alert volume trends, camera uptime. Use whenever the user asks for a "fleet report", "MSP report", "client summary", "weekly health across clients", "monthly rollup", a report spanning multiple orgs, or wants a shareable artifact about the state of their managed fleet. Produces markdown reports that can be emailed, pasted into a ticket, or committed to a knowledge base.
|
| argument-hint | [weekly|monthly|incident] [optional: client filter] |
| allowed-tools | Read, Bash, Write |
Cross-Client Reporting
Build shareable reports spanning every managed client org. Reports are generated in markdown; if the user wants PDF or HTML, recommend pandoc after generation.
Report templates
See references/report-templates.md for three templates:
- Weekly health — offline cameras, alert volume, door controller issues, battery-low sensors, per-client.
- Monthly executive — trend lines (alert volume week-over-week), license utilization, top 5 noisy clients, top 5 healthy clients.
- Incident rollup — one client, one window. Used after a notable event for an internal write-up.
Data sources
All data comes from the CLI (from the rhombus-partner-cli skill):
- Client list:
rhombus partner get-partner-clients-v2
- Cameras:
rhombus camera get-minimal-camera-state-list --partner-org <client>
- Doors:
rhombus door get-minimal-door-state-list --partner-org <client>
- Alerts:
rhombus alert recent --partner-org <client> --max 500 --after "7d ago"
- Licenses:
rhombus license get-license-state --partner-org <client>
For speed, loop over partner-clients-v2 and aggregate in memory before rendering.
Process
- Confirm scope — which report template? Which clients (default: all)? Which time window (default: last 7 days for weekly, last 30 for monthly)?
- Collect data — loop every client, running the queries above in parallel where possible (
xargs -P).
- Render — fill the appropriate template from
references/report-templates.md. Include:
- A headline summary (one sentence, one or two key numbers)
- Per-client table
- Callouts for outliers
- Recommended actions (ranked by impact)
- Output — print to stdout by default; if the user specifies an output path,
Write to that file.
Delegation
For the actual collection work, you can delegate to the rhombus-fleet-ops agent — it has an optimized loop and knows how to recover from per-client failures (e.g., a single client with auth issues shouldn't kill the whole report).
Output conventions
- File naming:
fleet-report-<weekly|monthly>-<YYYY-MM-DD>.md
- Formatting: headline numbers in the opening summary; detail tables below; trend arrows (↑/↓/→) for week-over-week or month-over-month comparisons.
- Confidential data: client names, camera names, and alert details. Warn the user before sharing outside their org.
Edge cases
- A client has 0 cameras. Skip from health tables; note in a footer.
- A client's API access is broken. Show "—" with a footnote; don't crash the report.
- You have >200 clients. Chunk the loop; report can take minutes. Offer a progress indicator.
- The user wants only specific clients. Respect their filter list rather than defaulting to "all".
Follow-up suggestions
After rendering:
- "Commit this to your shared Knowledge Base"
- "Email it to your customer success team"
- "Drill into [top-alerting-client] with
/rhombus-client-alerts <name> 7d"