Skip to content
Portfolio/Capabilities

CAPABILITY 26 / 30

The front door

A vague request gets scored, then asked two or three questions that actually change the answer.

IN PRODUCTION

This runs in client systems today. The sample below was regenerated for a fictional company.

Most bad output starts with a request nobody could have answered well. The front door scores the request against a rubric, shows what is missing, and asks the two or three questions from a register that would improve the answer most. It is not a prompt trick. It is declining to start work that cannot go well.

TRY IT / RUNS IN YOUR BROWSER

Use it here.

The mechanism is the production one. The company, the catalogue, and every figure are fictional.

The front door

Score the request before anyone starts writing

A vague request is the commonest cause of a bad output. This panel reads a request against a published rubric, shows what is missing and why it matters, then draws the two or three questions from a register of thirty that would change the answer most. Type anything. The scoring and the question choice run here, over the register, so an unusual request still gets a sensible score and sensible questions.

THE OUTPUT / GENERATED FOR APEX INSTRUMENTS

What comes out.

JSON
The question register, 30 questions in 7 groupsData · Grouped by what they resolve: audience 5, format 5, scope 5, source of truth 4, deadline 4, approval 4, confidentiality 3. Each question carries its prompt, three to five fixed options, the request kinds it applies to, an impact from a published three point scale, a sentence saying how much it changes the output, and for every option the clause it puts into the rewritten request. The rubric scores a dimension the register deliberately has no question for, and the register says which one and why.Open the file ↗
JSON
The rubric, 8 weighted dimensionsData · Audience 18, source of truth 16, scope 15, format 14, deadline 11, approval 10, example 9, confidentiality 7, summing to 100 so the score is already out of 100. Each dimension publishes what it looks for, why it matters, the sentence it says when it is missing, and the patterns that score it at full and half credit. Also carries the eight request kinds the classifier reads, the four score bands, and the selection formula the register is ranked by.Open the file ↗

THE MECHANISM / SHOW THE SHAPE, NOT THE SOURCE

How it works.

What a skill is
  1. 01Score it

    Audience, format, source, deadline, scope, example, approval, confidentiality.

  2. 02Show the gaps

    Each missing dimension is named, with what it would take to fill it.

  3. 03Ask from the register

    Questions come from a fixed register, not invented on the spot.

  4. 04Offer options

    Multiple choice, so answering takes seconds rather than a paragraph.

  5. 05Rewrite the request

    The enhanced version shows what changed and why.

WHAT IT READS

  • A scoring rubric with weighted dimensions
  • A register of questions with fixed options
  • The request, in whatever words arrived

WHAT IT CHECKS BEFORE IT SHIPS

  • The questions asked change with the request
  • Every question comes from the register
  • The enhanced request contains every answer given
  • A well-formed request is not interrogated for the sake of it
In production: a prompt enhancer front door that scores an incoming request and asks two or three multiple-choice questions from a register.

The mechanism runs inside client systems. Everything on this page was regenerated for Apex Instruments, a company that does not exist. Client names, source files, and internal wording stay private.

What it producesWhat is being built next