Agenda Time Boxer

Most meetings fail before anyone joins: ten topics, sixty minutes, no owners, no decision named. This skill builds an agenda backwards from the one outcome that must happen, allocates minutes the way a finance lead allocates budget, and cuts whole topics instead of shaving every item thin. Attendee time is treated as the scarcest resource in the company, because it is.

When to use this skill

• Someone asks for an agenda, a running order, or help structuring a meeting
• A topic list exists but exceeds the time available and something has to give
• A recurring meeting has drifted into status theater and needs rebuilding around decisions
• A workshop or planning session needs time boxes with owners and named outputs
• A meeting invite is about to go out with "Discussion" as its only agenda line

Instructions

1. Collect three inputs before doing anything: the total slot length, the single outcome that must happen for the meeting to have been worth holding, and the raw topic list.
2. Classify every topic as decision, discussion, or broadcast. Broadcasts — status updates, FYIs, anything one person says while the rest listen — move to a pre-read by default. Any broadcast that stays in the room needs a stated reason for being there.
3. Sequence decisions first, while attention is fresh; discussion second; logistics last. Never open a meeting with recaps.
4. Allocate boxes in five-minute units. Decision items get double your gut estimate, because the gut estimate assumes agreement. Boxes must sum to 90% of the slot — the remaining 10% is buffer, and the buffer is not negotiable.
5. Give every box an owner and a deliverable phrased as an outcome: "choose vendor A or B", not "discuss vendors". A box with no deliverable is a broadcast in disguise; see step 2.
6. Reserve the final three to five minutes for a wrap box: actions restated with owners and dates.
7. When topics do not fit, cut whole items from the bottom of the priority order and hand the requester the cut list. Ten shallow conversations are worth less than four finished ones.
8. Output the agenda plus the pre-read list, each pre-read with an owner and a send-by time.

Output format

A 45-minute decision meeting, done properly:

Meeting: Q3 vendor selection — Thu 10:00-10:45
Outcome: signed-off vendor choice and a migration start date
Pre-reads (sent by Wed 15:00): comparison memo (M. Osei), cost model (T. Vance)

| Time  | Box                                    | Owner    | Deliverable                  |
|-------|----------------------------------------|----------|------------------------------|
| 10:00 | Frame: criteria recap, no relitigating | M. Osei  | shared criteria (5m)         |
| 10:05 | Decision: vendor A vs vendor B         | T. Vance | vendor chosen (20m)          |
| 10:25 | Discussion: migration sequencing       | R. Kaur  | draft sequence + risks (12m) |
| 10:37 | Buffer                                 | chair    | absorb overruns (4m)         |
| 10:41 | Wrap: actions, owners, dates           | chair    | recorded actions (4m)        |

Guardrails

• Never produce an agenda whose boxes sum past the slot. An arithmetic failure here is a credibility failure everywhere else.
• One decision per box. Two decisions packed together means the second gets made in the hallway afterwards, by whoever lingers.
• If the requester insists on ten topics in thirty minutes, say plainly that the agenda will fail and propose a split or a pre-read cull. Do not quietly build the doomed version.
• Buffer is spent by the chair in the moment, never claimed in advance by the loudest topic owner.
• Pre-reads sent less than a day ahead do not count as pre-reads; the topic returns to the room and the agenda gets rebalanced honestly.