Skip to main content
Weekly intelligence brief that reads ALL evidence behind every spike, deal blocker, churn risk, and competitor mention — then synthesizes into a 40-line brief you can scan in 60 seconds. Separates what’s breaking (support) from what customers are asking for.

​
Usage

/closedloop:weekly-brief
No arguments needed. The skill automatically analyzes the last 7 days and compares against the week before.

​
What You Get

  • Executive summary — the story of the week in 4-6 lines, written after reading everything. Tells you what happened in plain language — not “1,423 insights from 258 customers” but “Rough week on the support side — two critical outages took down ticket sales and POS payments.”
  • Attention items — deal blockers, churn risks, and critical outages that need immediate action. Max 5 items, ranked by criticality.
  • What’s driving the numbers — for every notable change this week, an interpreted explanation of WHY, split by source. “Integrations up 54%: 180 from support — mostly sync failures. 68 from calls — operators asking for new connectors.”
  • Notable voices — the most vivid customer quotes from the week, with full conversation context from the actual call transcript
  • Competitive intelligence — which competitors were mentioned this week, by whom, and whether it’s a threat or opportunity

​
How It Works

The skill launches 4 parallel research agents:
  1. Numbers agent — gets volume, categories, feature areas, source split (support vs. calls), and week-over-week deltas
  2. Urgent items agent — finds every deal blocker and churn risk, reads full details for each
  3. Spike investigation agent — for every area that spiked, reads the actual insights to explain WHY, split by support vs. calls
  4. Competitive + voices agent — finds competitor mentions, loads conversation transcripts for the best quotes
After all agents complete, the skill assembles the brief — leading with the executive summary (written last, after reading everything).

​
Support vs. Calls Split

Support tickets are inherently negative — customers contact support when something’s broken. Sales and CS calls capture a broader range — feature requests, discovery, strategic discussions. Mixing them equally makes the brief read like “everything is on fire.” The weekly brief separates them:
  • From support — what’s breaking, what needs fixing
  • From calls — what customers are asking for, what to build next
This distinction tells you whether you have an operational health problem or a strategic product gap.

​
Quiet Weeks

When nothing spiked, nothing urgent, and no new themes emerged, the brief says so plainly — “Stable week” — and shows a shorter 25-line version with interesting individual quotes instead of fake urgency.

​
Data Quality

The skill silently cleans the data before counting:
  • Automated bot-generated insights (system error templates repeated identically) are excluded
  • Over-extracted calls (same speaker, same topic, multiple near-identical insights) are counted as one topic
  • Clean numbers are reported without mentioning data quality artifacts