Agent Skills
› mohitagw15856/pm-claude-skills
› weekly-review-ritual
weekly-review-ritual
GitHub建立30分钟周五复盘仪式，通过关闭本周循环、清空收集点、选定下周三大目标及反思问题，帮助用户掌控工作节奏，防止任务遗漏。
Trigger Scenarios
set up a weekly review
my weeks just happen to me
GTD-style review but lighter
I keep dropping threads between weeks
Install
npx skills add mohitagw15856/pm-claude-skills --skill weekly-review-ritual -g -y
SKILL.md
Frontmatter
{
"name": "weekly-review-ritual",
"description": "Install the weekly review that keeps work from managing you — the 30-minute Friday ritual: close the week's loops, sweep the capture points, choose next week's big three before the calendar chooses for you, and the two questions that compound. Use when asked set up a weekly review, my weeks just happen to me, GTD-style review but lighter, or I keep dropping threads between weeks. Produces the ritual's fixed agenda, the sweep checklist, the big-three selection, and the survival rules for busy weeks."
}
Weekly Review Ritual Skill
Without a weekly review, weeks are driven by whatever arrived loudest — the calendar fills by others' invitations, threads drop at the boundaries, and Monday starts in reactive mode by default. The review is 30 minutes, same slot weekly (Friday afternoon: close the week while it's warm, aim the next before it arrives), with a fixed agenda that never varies: close (the week's loops: done, stalled, dropped-on-purpose), sweep (every capture point emptied into the system), choose (next week's big three, calendar-checked), and the two compounding questions — what worked, what got dodged. The ritual's power is its boringness: same slot, same steps, every week including the bad ones.
What This Skill Produces
- The fixed agenda — the four phases with time boxes, tuned to the user's actual systems
- The sweep checklist — the user's real capture points (inbox, notes, chats, the desktop, the head) — enumerated so the sweep is mechanical
- The big-three method — next week's three outcomes chosen against the calendar's reality, blocked before Monday claims the space
- The survival rules — the 10-minute crisis version, and the never-skip-twice rule
Required Inputs
Ask for these if not provided:
- The existing systems — task list, calendar, notes; the review orchestrates what exists (email-triage-system, task-triage-matrix are its natural companions) — a review without a task system to sweep into needs that installed first
- The capture points, honestly — everywhere commitments accumulate: inboxes, chat mentions, meeting notes, the head (the head is always on the list)
- The week's shape — when the review can actually happen (Friday 3pm default; the slot must be real, recurring, and defended)
- The dropped-thread history — what tends to fall between weeks; the sweep checklist gets built around the actual leaks
Framework: The Ritual Rules
- Close before choosing: phase one walks the week's intended outcomes — done (acknowledge it: momentum is data), stalled (why? — the answer routes to unblock-protocol or renegotiation), and dropped-on-purpose (said explicitly, per the decision-journal honesty: a conscious drop is a decision; a silent one is a leak).
- The sweep is mechanical, from a list: each capture point opened and emptied — every loose commitment becomes a task, a calendar block, or a conscious no (email-to-tasks grain). The head-sweep closes it: two minutes, "what am I carrying that isn't written anywhere?" — the question that empties the 3am anxiety cache.
- The big three are outcomes, calendar-checked: three finishable outcomes for next week ("ship the draft to legal" not "work on the doc") — then held against the calendar's actual open hours (meeting-cost-meter reality: a 4-hour week can't carry three big outcomes; better to know Friday than Thursday) — and blocked into it (deep-work-blocking) before Monday's invitations arrive.
- The two questions compound: what worked this week? (do more of it — explicitly named) and what did I avoid? (the dodge list — usually the same item three weeks running, which is the review telling you something the task list can't). Thirty seconds each, written, because patterns emerge only in writing.
- Survival rules keep the streak: the crisis version is 10 minutes (sweep + big three only — closing waits), the never-skip-twice rule is absolute (one missed review is a busy week; two is the system dying), and the review reviews itself monthly: is the checklist still matching the real capture points?
Output Format
Weekly Review: [name] — [slot], 30 min
The Agenda
[Close (8 min): the week's outcomes → done/stalled/dropped-consciously · Sweep (12): the checklist below · Choose (8): big three vs. the calendar · The two questions (2), written]
The Sweep Checklist
[The user's actual capture points, in order · ending with the head-sweep question]
The Big Three (this week's)
[Three outcomes · the calendar check · the blocks placed]
Survival Rules
[The 10-minute crisis version · never-skip-twice · the monthly self-check]
Quality Checks
- The slot is real, recurring, and defended on the calendar
- The sweep list names every actual capture point including the head
- The big three are outcomes, checked against real open hours, and blocked
- Both questions get written answers
- The crisis version and never-skip-twice rule are explicit
Anti-Patterns
- Do not review without sweeping — a review that skips the capture points is a diary entry
- Do not choose five big things — three, calendar-checked; ambition unchecked by hours is next Friday's guilt
- Do not let drops happen silently — dropped-on-purpose is healthy; dropped-by-leak is the thing this ritual exists to end
- Do not skip twice — the second skip is the system's funeral, quietly
- Do not expand the ritual — 30 minutes is the ceiling; a 90-minute review gets skipped by week three
Version History
- 471c606 Current 2026-07-22 14:06
