23 — QA

Prove mechanical integrity, visual usability, genericity, and authority compliance before any presentation or release decision.

PARTIAL Partly built

Standing verdicts are HOLD, not PASS. The V46 architecture re-audit returned HOLD_FOR_GENERIC_FIXES; the 2026-09-04 spell pixel audit returned HOLD.

Deterministic Products 0 archived step 23 of 26

Where this output sits

Source evidence›Candidate evidence›Human decision›Audit acceptance›Release›Production

Source evidence, candidate evidence, human decision, audit acceptance, release and production are six separate things. This step produces the highlighted one and nothing further.

Scope

Runs the mechanical gates — token ownership, residual ink, background continuity, paragraph consistency, collisions, media bounds, table support, instrument state — and requires visual inspection at fit and at 300 DPI. Hash and token arithmetic alone may never pass a page.

This role is allowed to

Recompute all denominators
Challenge happy-path results with negative fixtures

How it works

fed by 22 ASSEMBLE

Inputs

Original pagesAll artifacts and productsPinned validator and browser/PDF environment

Deterministic Deterministic validators plus required human visual inspection and independent audit where policy requires.

Mechanical gates — token ownership, residual ink, background continuity, paragraph consistency, collisions
Media and table gates — currently unevaluable, because PIC and TABLE-PHYSICAL have no tool
Visual inspection at page fit and at representative 300-DPI crops
Verdict: PASS, or HOLD with stated reasons. Hash and token arithmetic alone may never pass a page

Must pass

All mandatory checks executed
Every page inspected at fit and representative 300-DPI crops
Console errors zero

passes →

Outputs

Token, sweep, background, typography, collision, media, table, instrument, visual and genericity reportsExplicit PASS/HOLD/FAIL per gate

hands off to PDF · DGTL/ARCH publication gate · Owner acceptance

fails →

HOLD

QA_INCOMPLETEQA_VISUAL_FAILQA_GENERICITY_FAILQA_AUTHORITY_FAIL

Nothing continues on a failed gate. Uncertainty becomes an explicit HOLD, and no later step may read an unanswered item as an accepted one.

Non-scope — what this role does NOT own

Passing a page on hashes or token counts alone
Declaring full-page accuracy from a sample
Authorising release

Explicitly forbidden

Treat hashes alone as quality
Self-authorize release
Ignore zero-denominator tests

Dependencies and position

Starts: After assembly; repeated after each compatible correction.
Previous step: 22 ASSEMBLE / GROUP
Next step: 24 PRINT / PDF
Hands off to: PDF, DGTL/ARCH publication gate, Owner acceptance

Exact inputs

Original pages
All artifacts and products
Pinned validator and browser/PDF environment

Exact outputs

Token, sweep, background, typography, collision, media, table, instrument, visual and genericity reports
Explicit PASS/HOLD/FAIL per gate

Performer and AI/ML boundary

Performer: Deterministic validators plus required human visual inspection and independent audit where policy requires.
Class: Deterministic
AI boundary: Auditor may assist, but evidence remains reproducible.

Training information

Not a trained component. No model, no weights, no training corpus. Behaviour is fixed by code and configuration, and identical inputs must produce identical outputs.

Deterministic validation and acceptance gates

All mandatory checks executed
Every page inspected at fit and representative 300-DPI crops
Console errors zero

HOLD and failure behaviour

Failure codes this step may emit:

QA_INCOMPLETE
QA_VISUAL_FAIL
QA_GENERICITY_FAIL
QA_AUTHORITY_FAIL

Uncertainty becomes an explicit HOLD. Omission never converts uncertainty into acceptance, and no downstream step may treat an unanswered item as an accepted one.

Downstream handoff

PDF, DGTL/ARCH publication gate, Owner acceptance

Active tool

Status: Partial
Run / inspect: python tools/qa_full_issue_v34.py / python tools/audit_v7.py
Input: assembled page packages
Output: audit.json + REPORT.md with an explicit verdict

Dependencies

Python 3.12 (NOT installed on this machine)

Why this is the active version

Mechanical QA tools exist and run, and an independent pixel audit was performed as recently as 2026-09-04. But the standing verdicts are HOLD, not PASS: the V46 architecture re-audit returned HOLD_FOR_GENERIC_FIXES, and the V004 spell audit returned HOLD as a list of real spelling errors. Several QA gates in the V8 contract cannot be evaluated at all because their upstream roles — PIC, TABLE-PHYSICAL, TABLE-ASSIGN — have no implementation.

Copied files — source receipt

File	Original SHA-256	Bytes	Copy
`qa_v34.py`	118930ebf79549c6…	8,409	verified
`qa_full_issue_v34.py`	1e2c942f0bcf2b4a…	6,265	verified
`audit_v7.py`	adff983905177c06…	11,631	verified
`validate_package.py`	d5c73279ec7b0706…	5,511	verified
`evidence/V004_SPELL_PIXEL_AUDIT_REPORT.md`	1062655048c932bc…	1,931	verified
`evidence/V004_SPELL_PIXEL_AUDIT_CLASSIFICATION.json`	aa01b3f2a7c6c3ae…	63,091	verified

6 file(s), all hash-verified against the original. Full detail in tool/SOURCE_RECEIPT.json. Copy-only: the historical source is never modified.

Archived versions

No superseded versions recorded for this role.

Known limitations

Standing verdicts are HOLD, not PASS. The V46 architecture re-audit returned HOLD_FOR_GENERIC_FIXES; the 2026-09-04 spell pixel audit returned HOLD.
Several gates cannot be evaluated at all because their upstream roles — PIC, TABLE-PHYSICAL, TABLE-ASSIGN — have no implementation.
Sample accuracy is not page accuracy. Earlier figures of 93.9 percent strict and 96.1 percent owner-relaxed were measured over 558 sampled words and must never be quoted as full-page accuracy.
A previous dashboard claim was withdrawn after direct image review found mislabelled pages and a classifier false positive.

Security and privacy

Operates on frozen local artifacts. No credentials required. No data leaves the machine.

Cost behaviour

No provider calls. No metered cost. Compute is local.

Copying a tool into this repository does not authorise running it, retraining it, calling a model, or processing a new issue. No paid call may be made without the owner's explicit authorisation and a hard cost cap.

Provenance

Registry: registry/roles.json (schema marsoom.e2t.roles.v1)
Derivation: 28 legacy roles - VER - HUMAN + REV = 27 active steps
Legacy role id: QA
Legacy source SHA-256: 6f5d5d7ed8869e45307424c2193f95ffab84abc44a6a188f5c7f98c2a48ec64d
Generated: 2026-09-04T22:13:12.433Z

Feedback and decisions

Feedback is recorded per source and never merged into an invented consensus. Where sources disagree, both positions stand and the owner decides.

Source	Events
Naser / owner Final authority. Overrides every other source.	none recorded
Codex / orchestrator Architecture and sequencing.	none recorded
Builder / designer Implementation reality and constraints.	none recorded
Independent reviewer Adversarial review of claims.	none recorded
Auditor Evidence verification against artifacts.	none recorded
Human REV reviewer Page-level truth from the review site.	none recorded

Recorded events

No feedback events recorded yet. The ledger exists and is append-only: feedback/FEEDBACK_LEDGER.jsonl.

Editing feedback is not possible: a change is a new event whose supersedes names the one it replaces, and the original stays exactly as written.

Generated from registry/roles.json by tools/build-reports.mjs. Do not hand-edit — edit the registry and rebuild.
Original page pixels are the visual authority. CAP owns the exact captured text.