How to run a critique in Thinkr

Critique is Thinkr's structured review pass. It reads a PRD the way a sharp staff PM would — checking problem framing, scope, evidence, and rigor — and hands back a scored verdict plus a list of specific gaps. You can run it two ways: standalone from the Critique page (a full saved report), or inside the PRD editor before you publish (findings land as inline comments). This guide covers both.

Before you start

You need something to critique: either a PRD you've generated in Thinkr, or a spec you can upload as a .pdf, .md, or .txt file. A critique runs against one document at a time.

To begin, open Critique from the sidebar. If you've never run one, you'll see the empty state — "No critiques yet" — with a New Critique → button. Otherwise click + New critique in the top-right. Both open the flow at /critique/new.

Step 1 — Pick what to critique

The new-critique screen is split: CONFIGURE on the left, CRITIQUE RESULTS on the right. On the left:

• Select PRD — search your workspace and pick a generated PRD from the dropdown ("Search and select a PRD…"), or
• Upload from file (.pdf, .md, .txt) — drop in a spec that doesn't live in Thinkr.

Then set two optional fields:

• Type (optional) — tells the critique what kind of work this is: General, 0 to 1 Feature, Core Refine, Growth/Experiment, Technical Debt, or Other. It tunes what the review weighs.
• Additional Notes (optional) — steer the focus, e.g. "Please focus heavily on edge cases."

When you're set, click Run Critique.

Step 2 — Let it run

The right panel switches from "Ready to critique" to a live progress view. The critique runs in numbered passes and takes around three minutes — leave the page open while it works. If you need to stop, hit Cancel.

Step 3 — Read the verdict and findings

When it finishes you land on the saved report. The left panel is an INDEX of every pass (click any row to jump to it); the right panel is the CRITIQUE RESULTS. At the top:

• Verdict — an overall score out of 5 with a one-line read on it:
  • ≥ 4 — "Strong PRD. Minor improvements possible."
  • 3–3.9 — "Good foundation but needs critical gaps addressed before building."
  • < 3 — "Significant gaps. Consider revisiting initiative fundamentals."
• Below that, each pass is a collapsible card — the specific gaps, risks, and conflicts the review surfaced, section by section.

Use Copy markdown (top-right) to pull the whole report into a doc or a message to your team.

Read the verdict last, findings first. The score tells you whether to ship; the findings tell you what to fix — and that's where the value is.

Step 4 — Act on it, then re-run

Take the findings back to your PRD, tighten the weak spots, and when you want a fresh read click Re-run Critique (bottom of the INDEX panel). You'll be asked to confirm — "Re-run this critique?" — and the previous report stays saved, so you can compare before and after.

Critique inside the PRD editor

You don't have to leave the editor to get this. On a draft's Review tab you'll see "Run a critique to catch gaps before you publish." That opens the Critique panel — the same engine — but instead of a standalone report, the gaps it finds are added straight to your Findings tab as inline comments you can resolve one by one. From the panel you can also View full report (it opens the standalone critique in a new tab) or Re-run.

So: use the editor critique to clean up a draft in place; use the standalone Critique page when you want a saved, shareable report — or to review a spec that isn't a Thinkr PRD.

That's the loop

Pick a PRD, run the critique, read the verdict, fix the findings, re-run. One critique before publishing is the fastest way to catch the missing non-goal or soft metric a reviewer would otherwise send back to you.

Ready to try it? Open Thinkr and click Critique.