rate

A structured quality read used to decide whether a piece is ready. Part of Review. It evaluates only — it does not rewrite.

The final score is computed, not chosen: each of the six dimensions is scored against fixed rubric anchors, then combined with the VPM subscore through a fixed weight profile (the weights depend on whether the piece is technical, opinion, or narrative), and the calculation is shown in the output. The dimension judgments themselves are still the model’s reading — see the note below. A deduction ledger names the observable defect that kept each dimension from the next level — and for your own drafts, recommendations (zero to three) may only be derived from that ledger. “No material score-improving changes identified” is a valid, expected result for a finished piece.

Quality scores are intrinsic: your config never moves them. Config (purpose.md, buckets.md, expertise.md, tonality.md) personalises only the strategic layer — the four goal ratings, best-fit platform, and publishing readiness.

When to use it

When you need a structured quality assessment — your own draft before publishing, someone else’s article before reading/watching, or two versions of the same piece being compared. Signals: “is this worth my time”, “is this ready to ship”, “how does this score”, or any ask for VPM (value per minute) or quality dimensions.

When not to use it

• For rewriting or improving the piece → use improve-writing.
• For flagging structural craft issues against the checklist → use diagnose.

Guidance, not ground truth

The score reflects the framework’s judgment, not an objective measure of quality. Use it to focus attention, then make the call yourself. See Known Limitations.

Run it once per session

The rating is poisoned by context when run repeatedly in the same conversation — earlier ratings and discussion bias later ones. Run it once per fresh session/context for a clean read. As a rule of thumb a piece scoring above ~80 is usually fine. The deduction ledger and the zero-recommendation rule exist to stop the evaluator from escalating nitpicks into “issues” — but a fresh context is still the cleaner instrument.

Example

Rate this draft — is it ready to ship?