The pre-mortem: find how your plan fails before it does
Hands the model a plan and asks it to write the post-mortem for a failure that hasn't happened yet. Surfaces the risks your team is too invested to name out loud.
Turns a week of scattered notes into an honest account of where the time went, what moved, and the one thing worth protecting next week.
Here is a raw record of my week. Help me see it honestly. RAW RECORD: {{RAW}} WHAT WAS SUPPOSED TO MATTER THIS QUARTER: {{GOALS}} Write: **Where the time went.** Rough percentages across categories you infer from the record. Name the largest category plainly, even if it's meetings or firefighting. **Against the goals.** For each stated goal: moved / stalled / not touched, and the specific evidence from the record. Do not be generous with "moved". **The gap.** One paragraph on the distance between what I said mattered and where the hours actually went. No pep talk. If they matched, say so — that's worth knowing too. **Three things that worked.** Specific, from the record, worth repeating. **One thing to protect next week.** A single commitment, with the specific hours blocked and the specific thing you'd say no to in order to keep it. Just one. Rules: - Use only what's in the record. If it isn't there, it didn't happen — say the record is thin rather than filling gaps. - No motivational language. No "great job on". Flat and factual. - If the week was genuinely bad, say that in one sentence and move on.
Tools for this prompt
Where the input below comes from. The ones marked needed change the output a lot — fallbacks for each are further down.
This prompt is only as good as what you feed it. Web search covers anything public — these are the inputs it can't reach.
What actually closed, as opposed to what you remember. Memory over-weights the interesting hours.
Don't have it: Ask the user to paste their list; note it may be less complete than the tracker.
Somewhere to keep twelve weeks of these, which is where the pattern shows up.
Don't have it: Output the document as markdown for the user to file wherever they keep things.
Only tools that web search cannot replace are listed. Anything public — reviews, job ads, papers, salary data — an agent can already fetch, so it is not listed here. No paid placements or affiliate links; if that changes, this line will say so.
A prompt is a skill with the serial numbers filed off. Install it once and your agent reaches for it on its own.
Paste this into ChatGPT, Claude Code, Codex, Cursor, or whatever you use:
Install the weekly-review-that-actually-sticks skill globally from https://promptsbuddy.com/prompts/weekly-review-that-actually-sticks/skill.md.claude/skills/weekly-review-that-actually-sticks/SKILL.md
---
name: weekly-review-that-actually-sticks
description: "Turns a week of scattered notes into an honest account of where the time went, what moved, and the one thing worth protecting next week. Use when the user asks for help with career tasks like productivity, reflection, career."
license: CC-BY-4.0
metadata:
source: https://promptsbuddy.com/prompts/weekly-review-that-actually-sticks
author: "Rony"
version: "1.1"
---
# The weekly review that survives a bad week
Turns a week of scattered notes into an honest account of where the time went, what moved, and the one thing worth protecting next week.
## Inputs to collect first
Ask the user for anything below that they have not already given you. Do not
invent values for these.
- `RAW` — Calendar exports, todo dumps, notes, commit messages, Slack you starred. Messy is fine. (e.g. Mon: 4 meetings. Shipped the export fix. Tue: interviews all day...)
- `GOALS` — The two or three things that were supposed to matter this quarter. (e.g. Ship self-serve onboarding. Hire one senior engineer. Stop being the bottleneck on reviews.)
## Tools this works with
Do not require any of these. If one is unavailable, use the stated fallback and
say which input is missing rather than inventing it.
- **Linear** — What actually closed, as opposed to what you remember. Memory over-weights the interesting hours.
- Without it: Ask the user to paste their list; note it may be less complete than the tracker.
- https://linear.app/
- **Notion** — Somewhere to keep twelve weeks of these, which is where the pattern shows up. If its MCP server is connected, fetch this yourself.
- Without it: Output the document as markdown for the user to file wherever they keep things.
- https://www.notion.com/
## Instructions
Here is a raw record of my week. Help me see it honestly.
RAW RECORD:
{{RAW}}
WHAT WAS SUPPOSED TO MATTER THIS QUARTER:
{{GOALS}}
Write:
**Where the time went.** Rough percentages across categories you infer from the
record. Name the largest category plainly, even if it's meetings or firefighting.
**Against the goals.** For each stated goal: moved / stalled / not touched, and
the specific evidence from the record. Do not be generous with "moved".
**The gap.** One paragraph on the distance between what I said mattered and
where the hours actually went. No pep talk. If they matched, say so — that's
worth knowing too.
**Three things that worked.** Specific, from the record, worth repeating.
**One thing to protect next week.** A single commitment, with the specific hours
blocked and the specific thing you'd say no to in order to keep it. Just one.
Rules:
- Use only what's in the record. If it isn't there, it didn't happen — say the
record is thin rather than filling gaps.
- No motivational language. No "great job on". Flat and factual.
- If the week was genuinely bad, say that in one sentence and move on.
## Notes from people who use this
- Feed it your calendar export, not your memory. Memory over-weights the interesting hours.
- Keep the outputs in one file. Twelve weeks of 'the gap' paragraphs is the most useful career document you'll own.
- 'One thing to protect' works because it's one. Three commitments is a wish list.
---
The weekly review that survives a bad week · by Rony · v1.1
From PromptsBuddy — https://promptsbuddy.com/prompts/weekly-review-that-actually-sticks
Licensed CC BY 4.0.
Every prompt is also available at /prompts/weekly-review-that-actually-sticks/skill.md — see all install options.
Reflection prompts usually come back warm — you did great, be kind to yourself. That produces a pleasant five minutes and no behaviour change. The instruction to drop motivational language leaves the numbers standing on their own, and the numbers are usually the point: you said hiring mattered and spent four percent of your week on it.
Working from a calendar export instead of recall removes the flattering edit memory applies by default. Commit messages, starred messages, and sent mail are all good raw material — anything with timestamps you didn't curate.
One week's review is mildly interesting. A quarter of them, read at once, shows a pattern you cannot see from inside any single week — which is the actual reason to keep doing it.
Hands the model a plan and asks it to write the post-mortem for a failure that hasn't happened yet. Surfaces the risks your team is too invested to name out loud.