exec-weekly-agenda-generator

name	exec-weekly-agenda-generator
description	Synthesize the past week across meetings, chat, email, and calendar to generate a ranked list of 5-10 topics for the weekly leadership team meeting. Use when the user asks to prep for their leadership meeting, generate meeting topics, build the agenda, or says "prep me for Monday."

Weekly Leadership Agenda Generator

Surface the highest-value topics from the past week for the weekly leadership team meeting. The goal is to replace memory-based prep with a systematic sweep of every signal source, so nothing falls through the cracks.

This is not a meeting-minutes generator or a status report. It is a decision-and-discussion agenda optimized for:

• topics that need the full leadership team (not 1:1s)
• items that are specific and traceable to this week's activity
• forced ranking so the meeting stays tight

Setup

Fill in once (or let the agent ask):

• [LEADERSHIP TEAM] — a roster of who's in the room and the domain each owns (e.g., "Alex — Product/Eng, Sam — Revenue, Priya — People, Jordan — Finance"). Used to tag which domains each topic touches and to route single-domain items to 1:1s.
• [MEETING NOTES TOOL] — where meeting notes/transcripts live (Granola, Fathom, Otter, Fireflies, etc.)
• [EMAIL] — your sent-mail, as a signal source
• [CHAT] — your team chat (Slack, Teams), as a signal source
• [AI CHAT HISTORY] (optional) — transcripts of your own AI/agent sessions from the week, if you keep them. Often where the deepest thinking happens.

The skill degrades gracefully: if a source isn't available, flag it and keep going with the rest.

When to Use

• The evening before or the morning of the weekly leadership meeting
• When the user says "prep for the leadership meeting," "what should I raise," "build the agenda," or "prep me for Monday"

Workflow

Copy this checklist and work through it:

Weekly Leadership Agenda Progress
- [ ] Step 1: Find last week's leadership meeting notes
- [ ] Step 2: Pull meetings from the past week
- [ ] Step 3: Pull sent email from the past week
- [ ] Step 4: Pull chat activity from the past week
- [ ] Step 5: Scan AI chat history from the past week (optional)
- [ ] Step 6: Extract candidate topics from all sources
- [ ] Step 7: Filter, score, and rank the top 5-10 topics
- [ ] Step 8: Format and present the agenda

Step 1: Find Last Week's Leadership Meeting Notes

From [MEETING NOTES TOOL], find the most recent prior leadership meeting. Extract:

• Decisions made — done; don't re-raise unless implementation stalled
• Action items assigned — carry-forward candidates
• Open threads — discussions that ended without resolution or were deferred
• Topics tabled — anything punted due to time

Label all carry-forward items clearly in the final output. They get automatic inclusion in the candidate list.

Step 2: Pull Meetings From the Past Week

Pull summaries for the week's meetings. For meetings with leadership-relevant signal, go into the transcript for detail.

Look for:

• Decisions that need leadership ratification or awareness
• Cross-functional issues that surfaced in multiple meetings
• Blockers or risks that require leadership intervention
• Strategic shifts or new information that changes priorities
• Alignment gaps (disagreements, unclear ownership, competing priorities)
• Personnel or org issues that need leadership awareness
• Wins worth celebrating

Skip: routine 1:1 status updates, scheduling logistics, single-domain topics already handled.

Step 3: Pull Sent Email

Scan sent mail from the past 7 days. Look for:

• Emails escalating or flagging an issue to multiple leaders
• External communications leadership should know about (board, investors, key customers, press)
• Strategic decisions communicated that need broader alignment
• Replies signaling a change in direction or priority

Skip: scheduling, short acknowledgments, routine approvals.

Step 4: Pull Chat Activity

Fetch recent messages the user sent or participated in. Focus on substantive threads, not reactions or one-liners.

Look for:

• Cross-functional debates or escalations
• Threads where the user weighed in on a product, strategy, or org question
• Threads involving multiple leaders (strong signal for full-team relevance)
• Urgent issues raised in chat that lack a clear owner or resolution
• FYI-level updates the leadership group should be aware of

Skip: social channels, short acknowledgments, HR/legal-sensitive threads.

Step 5: Scan AI Chat History (Optional)

If the user keeps transcripts of their AI/agent sessions, scan the last 7 days for:

• Strategic analysis or decision-framing the user worked through
• Data pulls or analyses that revealed something leadership should see
• Issues the user spent significant time on (time invested = importance signal)
• Hiring, org design, or people topics processed

This is often where the deepest thinking happens. Mine it carefully.

Step 6: Extract Candidate Topics

Cast a wide net. For each source, pull out anything that could be a leadership topic. Capture:

• The raw issue or signal
• Which source(s) it came from
• Which domain(s) it touches (map against [LEADERSHIP TEAM])
• Whether it's a carry-forward from last week

Aim for 15-25 raw candidates before filtering.

Step 7: Filter, Score, and Rank

Filter First

Remove any topic that:

• Only involves one leader's domain and can be handled in a 1:1. Not worth full-team time.
• Is generic or standing (e.g., "review metrics"). Every topic must trace to something specific that happened this week.
• Is already resolved. Don't re-raise a made-and-communicated decision unless implementation is off track.
• Is HR/legal-sensitive and shouldn't go to the full group. Flag separately for offline handling.

Score Remaining Topics (1-5 each)

• Cross-functional reach: How many domains does this touch? 2+ domains score highest.
• Urgency: Is there a deadline? Topics with deadlines in the next few days get a 5 and a deadline marker.
• Impact: How much does this matter for direction, revenue, product, or team? Strategic and customer-impacting issues score highest.
• Unresolved tension: Active disagreement, unclear ownership, competing priorities? These are the highest-leverage topics because only the full team can resolve them.

Composite = reach + urgency + impact + unresolved tension.

Rank and Cut

• Keep the top 5-10.
• If more than 10 survive, force-rank and cut.
• Still-open carry-forwards from last week get automatic inclusion (but are still ranked).
• Cross-cutting signals (topics that appeared in 2+ separate conversations) get a ranking boost.

Step 8: Format and Present

# Weekly Leadership Agenda — [date]

Sources scanned:
- [X] meetings reviewed
- [X] sent emails scanned
- [X] chat threads reviewed
- [X] AI chat sessions scanned
- Last week's meeting: [found/not found], [X] carry-forwards identified

---

## 1. [Topic Title]

**Treatment:** discuss | inform | decide | follow-up
**Involves:** [Leader name(s) / domain(s)]
**Source(s):** [Meeting name, chat thread, email, transcript — be specific]
[CARRY-FORWARD] — if from last week
[DEADLINE: date] — if time-sensitive
[CROSS-CUTTING] — if surfaced in 2+ conversations

**Why it matters / why now:**
[2-3 sentences. Be specific about what changed or what is at risk.]

**Suggested framing:**
[What to say or ask to open the topic. 1-3 sentences in the user's voice.]

---

[Repeat for each topic]

---

## Below the Cut
[Topics that didn't make the top 10, one line each, so the user can override the ranking]

## 1:1 Candidates
[Important but single-domain topics — better handled in a 1:1. List with the relevant leader so they can be routed.]

Treatment Definitions

• discuss: Needs group input. The user opens the topic and facilitates.
• inform: FYI only. Context shared, no discussion unless someone flags a concern.
• decide: Needs a call. The user frames the decision and the team makes it in the meeting.
• follow-up: Checking on a prior commitment. Quick status, not a new discussion.

Guardrails

• No generic standing items. "Review metrics" is not a topic. "Q2 pipeline is tracking 15% below plan and the team disagrees on whether to adjust targets" is a topic.
• No single-domain items in the main list. Route them to 1:1 Candidates.
• Max 10 topics. If you have 12, cut harder. A short agenda with real discussion beats a packed one that becomes a status report.
• Every topic must cite its source. If you can't point to a meeting, thread, email, or transcript, it doesn't belong.
• Carry-forwards and cross-cutting signals are clearly labeled.
• Deadline items are flagged.
• No confidential HR/legal items in the main list.

Edge Cases

• Light week: Still produce 3-5 topics; lean on carry-forwards and chat. A quiet week may itself be worth noting ("no escalations this week — is that good, or are we not hearing about problems?").
• A source is unavailable: Flag it explicitly and proceed with the rest. Don't silently skip.
• No carry-forwards from last week: Note it. Either last week was unusually clean, or notes weren't captured.
• Too many high-priority topics: If 15+ clear the cutoff, group related items into compound topics (e.g., "Q2 revenue risks" combining pipeline + pricing + churn) with sub-bullets.

Anti-Patterns

• Don't generate a status report. This is an agenda for decisions and discussion.
• Don't include already-resolved topics unless implementation is stalled or the resolution needs ratification.
• Don't pad the list to hit 10. Five strong topics beat ten mediocre ones.
• Don't skip a source silently — flag it and explain.
• Don't invent topics. Every item must trace to a real signal from the week.
• Don't include detailed minutes or long summaries. Keep each topic tight: title, why, framing, who, source.