Club Benefits
Most weekly reviews are a little too kind. You look back at what you did, feel reasonably productive, jot down a few intentions for next week and close the notebook. It feels reflective, but nothing really changes because nothing was examined very closely.
In other words, you're reviewing your week the same way you'd review a friend's work — generously, charitably and with the benefit of the doubt. A few months ago I realized I needed to stop this cycle. I turned to Claude to try something different, and now after reaching more goals and finding holes in my workflow, it's the only way I run my weekly review.
Article continues belowWhat is the MIRROR system?
MIRROR stands for: Map, Interrogate, Reality-check, Reveal, Optimize, Reset.
Each stage is a distinct prompt. You run them in sequence in a single Claude conversation, building on each previous response. By the end you have something more honest than a journal entry and more actionable than a mood board. You have a clear-eyed account of your week — work, habits, creative output, personal life — and a specific plan for the next one.
Stage 1: Map
Purpose: This stage helps to create an honest inventory of your week before you start evaluating it.
The biggest mistake people make in weekly reviews is jumping straight to evaluation before they've actually mapped what happened. Memory is selective. You remember the wins and the dramatic failures. The quiet drift — the hours that disappeared while scrolling TikTok, the things you said you'd do and didn't — gets edited out before you even start.
The prompt: "I'm going to give you a brain dump of my week — everything I did, worked on, felt and experienced. It will be messy. Don't organize it yet. Just read it and then give me back a clean inventory: what I actually did (work), what I intended to do but didn't, what happened in my personal life, and how I seemed to be spending my energy based on what I've described. [Paste your brain dump here]"
Claude Sonnet 4.6, the default and free tier model, features significantly improved long-context reasoning and agent planning — which means it's genuinely good at reading a long, unstructured brain dump and pulling out a coherent picture without losing details. What you get back at this stage usually contains at least one thing you'd already forgotten about your week. That's the point.
Stage 2: Interrogate
Purpose: Here is where Claude gets to ask the questions about your week that you wouldn't think to ask yourself. This is the stage that makes MIRROR different from every other review framework. You're not just asking Claude to summarize — you're explicitly asking it to probe.
The prompt: "Based on the inventory you just built, I want you to interrogate this week. Specifically: what did I say was a priority that my actions suggest wasn't? Where did I spend time that I didn't mention as important? What pattern do you see across the things I didn't finish? Ask me three follow-up questions about the parts of my week that seem underexplained. Don't be polite — be precise."
The three follow-up questions Claude asks at the end of this stage are usually the most valuable part of the whole review. Answer them honestly before moving to Stage 3. Nobody has to see your answers.
Stage 3: Reality check
Purpose: Now is the time to compare what you told yourself about your week to what the evidence suggests. Everyone has a story they tell about themselves, such as "I'm so productive!" But the Reality check stage asks Claude to look at the evidence from your actual week and tell you where the story holds up and where it doesn't. Busy doesn't mean productive.
The prompt: "Here is how I would describe myself and my values: [two or three sentences about who you think you are and what you prioritize]. Based on everything you know about my week from the previous stages, where did my behavior match that description and where did it contradict it? Be specific — use examples from my actual week, not generalities."
The instruction "use examples from my actual week" is critical. Without it, Claude will give you a thoughtful but generic answer. With it, you get something that feels uncomfortably specific — because it is. Claude Sonnet 4.6's expanded context window lets it ingest large amounts of information and reason effectively across that context, which means by Stage 3 it has a detailed enough picture of your week to back up every observation with evidence.
Stage 4: Reveal
Purpose: Here's where everything comes to the surface. Claude will help you discover what you're not seeing. It's this stage that often surprises me the most. . After three stages of honest inventory, interrogation and reality-checking, there's usually a signal in the noise — a theme, a pattern, a tension that keeps appearing. The Reveal stage asks Claude to name it.
The prompt: "Having looked at my week across all these dimensions — what I did, what I avoided, where my behavior matched my values and where it didn't — what do you think my week is trying to tell me? Not what I should do differently. Not advice. Just: what is the signal in this week that I'm probably not paying enough attention to? State it plainly."
"State it plainly" stops Claude from hedging. The reveal is only useful if it's specific enough to be uncomfortable. In my experience running this stage over several months, it has surfaced the same theme three weeks in a row before I finally acknowledged it. That's not a coincidence — it's the point. The things we most need to hear are usually the things we keep creating evidence for and then explaining away.
Stage 5: Optimize
Purpose: This stage turns everything you've learned into a specific, realistic plan for next week.
Most review frameworks end at reflection. MIRROR doesn't. The Optimize stage takes everything from the first four stages and converts it into concrete decisions — not intentions, not goals, but decisions.
The prompt: "Based on everything from this review — the gaps between intention and action, the reality-check findings, and the signal you identified — give me next week's plan. Not a to-do list. Three specific behavioral commitments that address what this week revealed, one thing I should explicitly stop doing, and one thing I should protect at all costs. Make it realistic based on what you know about how I actually operate, not how I'd like to operate."
"Make it realistic based on what you know about how I actually operate" is the key phrase. Claude Sonnet 4.6 is materially better at long-horizon planning — the kind of work where the model has to keep objectives, constraints and progress consistent over time — and by Stage 5 it knows enough about the gap between your intentions and your actual behavior to plan for who you are rather than who you wish you were. That's a plan you might actually follow.
Stage 6: Reset
Purpose: Now is the time to close the loop and get one clear sentence to carry into the next week.
This final stage is short by design. You've done a lot of hard looking. The Reset gives you something simple and forward-facing to leave with.
The prompt: "Given everything from this review, give me one sentence — not a to-do, not a goal, but a principle or intention — that I should keep in mind this week. Something I can come back to on Thursday when things get difficult. Make it specific to me, not generic."
The sentence Claude gives you at this stage is almost always more useful than anything you'd write yourself, precisely because it's based on evidence rather than aspiration. Mine three weeks ago was: "Finishing one thing matters more than starting three." I didn't come up with that. I wouldn't have, because it implicitly admits something I don't like admitting. But it was exactly right.
What MIRROR keeps finding that I ignore
The gap between what I say matters and what my time says matters. Every week. In different areas — sometimes work, sometimes self care, sometimes parenting — but always the same structural gap. I say one thing is a priority and then I spend Tuesday doing something else entirely, and by Sunday I've constructed a perfectly reasonable explanation for why that was actually fine.
MIRROR doesn't accept the explanation. It looks at the evidence, names the pattern and asks me what I'm going to do about it next week. Most of the time I already know. The problem was never not knowing what to do. It was the willingness to look directly at it without flinching.
That's what a real mirror is for.
The takeaway
Getting started with the MIRROR system is easy. Just open a new conversation in Claude on claude.ai — the free Sonnet 4.6 tier handles all six stages without any issues. Block 30 minutes on Friday evening. Start with a messy brain dump in Stage 1 and don't edit it before you paste it in. The mess is the point. Claude will sort it out.
Run all six stages in the same conversation so Claude maintains context across the whole review. Don't skip Stage 2. That's where most of the value lives. Give it a try and let me know what you think in the comments.
Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds.
More from Tom's Guide
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.
