Skip to content
HOW WE JUDGE THE STANDARD

A credential is worth what it refuses.

An assessment that passes everyone is selling attendance. These are the rules this platform enforces in code, with the file that enforces each one, so the claim can be checked rather than believed.

  1. 01Work that was not judged is never recorded as failed.
  2. 02A heuristic cannot verify anything.
  3. 03Only a live review can clear a requirement.

All 6 rules, with their enforcement →

THE ASSESSMENT · AUTHORED FACTS
Rules enforced in code
6
Source files that enforce them
3
Things this page refuses to claim
4
Capability dimensions carried
6
Global scores or percentiles
0
Ranking surfaces in the product
0

Each count is the length of a list published on this page. The zeroes are the point: none of those surfaces exists to be switched off.

THE RULES

Six, and each one costs us something.

A rule that never refuses anyone is not a rule. Each of these makes the platform say less than it could.

  1. 01

    Work that was not judged is never recorded as failed.

    When no review provider is reachable, a keyword and length check runs so the submission is not lost. It produces a number and no verdict. The record reads NO VERDICT — never NOT PASSED, which would imply the work was read and found short.

    ENFORCED IN

    src/lib/review-provenance.ts

  2. 02

    A heuristic cannot verify anything.

    That offline check has its pass flag forced false by construction. It cannot mark an artifact verified, cannot close a node, and cannot clear a role floor. It is recorded because the submission happened, not because it was assessed.

    ENFORCED IN

    src/lib/review-provenance.ts · src/lib/credentials.ts

  3. 03

    Only a live review can clear a requirement.

    Offline values stay visible as diagnostic observations. A role floor is cleared only by a live-reviewed assessment, which is deliberately stricter than showing the latest numbers.

    ENFORCED IN

    compareRoleDemands in src/lib/credentials.ts

  4. 04

    A result is a profile, not a number, and nobody is ranked.

    Capability is carried as six dimensions with a floor per role. There is no global score, no percentile, no league table and no comparison between people. None exists to disable — the platform has no ranking surface.

    ENFORCED IN

    src/lib/credentials.ts

  5. 05

    Machine review and human review never convert into each other.

    A model verdict is a machine verdict on submitted work, not a human judgment and not a statement about the person. Expert review is recorded separately and no process promotes one into the other.

    ENFORCED IN

    src/lib/review-provenance.ts

  6. 06

    A reviewer is named, or the record says they are not.

    A practitioner is named on a record only when the roster records them as confirmed and cleared for publication. Until then the record publishes the standard that was applied rather than the person who applied it — it does not quietly omit the question.

    ENFORCED IN

    src/lib/review-provenance.ts

WHAT WE DO NOT CLAIM
  • No assessment here verifies a person’s identity. Nothing is biometric and nothing proves who typed.
  • No result predicts job performance, and none is an employer screen. A verdict is about submitted work.
  • No employer has validated this platform’s judgements, and no placement or affiliation is implied by any record it issues.
  • A completed course is not advancement. Reading and rehearsal prepare someone for production work; they never substitute for it.

The vocabulary the assessment is written in is published as the Capability Atlas.