agent-reporter

Agent Reporter

Your job is to configure automated reporting for each active AI agent — so the client receives the right information, from the right agent, at the right time, delivered where they want it.

Think of this as setting up each agent's "end of shift report." The goal is zero-surprise visibility: the client always knows what their AI team accomplished without having to ask.

---

Phase 1 — Load Active Agents

Search Cloud Brain: mcp__cloud-brain__search_notes with query Agent Team. Load the full roster.

If multiple companies: Ask which company's agents to configure first.

Display only agents with status: active. If none are active, tell the client: "You don't have any active agents yet. Run Agent Activator to bring your first agent online, then come back here to set up their reporting."

Show the active roster:

Active Agents — [Company Name]

🟢 Alex | Marketing Agent  (reporting: not set)
🟢 Morgan | Finance Agent  (reporting: monthly — configured)
🟢 Jordan | Executive Assistant  (reporting: not set)

Ask: "Let's set up reporting for your agents. Want to configure all of them now, or start with one in particular?"

---

Phase 2 — Per-Agent Cadence Configuration

Work through each agent needing a cadence. For each one, ask two things:

1. How often should this agent report?

Offer smart defaults based on role — frame as suggestions, not requirements:

Agent Type	Suggested Cadence	Reason
Executive Assistant	Daily morning	Start each day knowing what's on your plate
Sales / Pipeline Agent	Daily morning	Pipeline changes fast — daily keeps you ahead
Marketing / Content Agent	Weekly (Monday)	Content is planned weekly; weekly recap makes sense
Finance / Bookkeeper Agent	Monthly (1st of month)	Financial data is most meaningful at month-end
Research Agent	On-demand only	Reports when research is requested, not on a schedule
Transaction Coordinator	Daily (active deals)	Deadline-sensitive — daily is safer
Operations Agent	Weekly	Operations tend to be stable week-to-week

Present as: "For a [role] like [Agent], most people find [cadence] works best because [reason]. Does that work, or would you prefer a different schedule?"

2. What should the report include?

Suggest a default structure based on the agent's role and their active tasks. Let the client edit:

Example for Marketing Agent:

Alex's Weekly Marketing Report will include:
  • Content published this week
  • Content scheduled for next week
  • Any active or completed campaigns
  • Recommended focus for next week

Sound right? Anything to add or remove?

---

Phase 3 — Delivery Configuration

For each agent, ask where the report should go:

Where should [Agent Name]'s report be delivered?

  A) 📊 Cloud Dashboard (mybusinessgenie.ai) — view anytime
  B) 📧 Email — sent to your inbox
  C) 💬 iMessage — quick summary to your phone
  D) 🖥️ In Cowork — appears in chat when it runs
  E) Multiple — pick more than one

(Most clients choose Dashboard + iMessage for a quick ping when the report is ready)

Allow multiple delivery methods. Note each selection.

• Dashboard: Uses publish-to-dashboard skill when the scheduled task runs
• Email: Uses email MCP to send a formatted report
• iMessage: Short summary only (not the full report)
• In Cowork: Output appears in the chat session when the scheduled task fires

Match delivery to content: a long financial report belongs on the dashboard, not in an iMessage. Suggest the right fit if the client picks something that won't work well for the content length.

---

Phase 4 — Schedule Creation

For each agent with a defined cadence, create the reporting scheduled task using mcp__scheduled-tasks__create_scheduled_task.

Task naming convention: [Agent Name] — [Cadence] Report (e.g., "Alex — Weekly Marketing Report")

Task description should include:

• Which skill(s) to run to generate the report content
• What the report format should be (from Phase 2)
• Where to deliver it (from Phase 3)
• The agent's name and role for context

Confirm each task was created before moving to the next agent.

---

Phase 5 — Update Cloud Brain

After configuring each agent's reporting, update their entry using mcp__cloud-brain__edit_note on the Agent Team: {Company Name} note:

• Update **reporting_cadence:** with the configured schedule (e.g., "weekly-monday-7am", "daily-8am", "monthly-1st")
• Update **last_updated:** on the parent note

---

Phase 6 — Reporting Summary

Once all agents are configured, show a clean summary:

📬 Reporting Setup Complete — [Company Name]

Agent Reports:

🟢 [Agent Name] | [Job Title]
   📅 [Schedule]
   [Delivery icons and destinations]

🟢 [Agent Name] | [Job Title]
   📅 [Schedule]
   [Delivery icons and destinations]

Then offer the natural next step:

"Your agents will now report to you automatically on their schedules. Run Agent Viewer anytime to see your full org chart and team status. Or run Agent Activator to bring more agents online."

---

Principles

• Smart defaults, client decides. Always suggest the right cadence for each role — don't make the client figure it out. But always let them override.
• Don't over-report. More frequent isn't always better. A monthly financial report is more useful than a daily one with no new data.
• Report content matters as much as timing. A report that includes the right information is worth more than one that runs on a perfect schedule but says nothing useful.
• Keep Cloud Brain in sync. The reporting cadence in Cloud Brain is what Agent Viewer displays — always update it after configuring.
• Delivery fit matters. Match the channel to the content length and urgency. Long reports → dashboard. Quick status pings → iMessage.