# STACK B — CANDIDATE TEST MATRIX

Version: `0.1.0-STACK-B-TESTS`  
Status: `CANDIDATE / NON_RUNTIME / NO_AUTHORITY_EFFECT`  
Scope: static and structural validation of the first Stack B candidate slice

## 1. Validation command

From the AGLA repository root:

```powershell
powershell.exe -NoProfile -ExecutionPolicy Bypass -File "AGLA_ArsGeneralisLlullApplied\_01_INTEGRATION_CANDIDATES\STACK_B\VALIDATE_STACK_B_CANDIDATE.ps1"
```

The command is read-only. It validates the candidate family, its JSON schema and fixtures, and the absence of canonical registration.

## 2. Positive baseline

The validator requires:

1. all five class artifacts plus local index, source register, gap analysis, schema, and fixtures;
2. the immutable fixture identity `HUMAN_OPERATUM_REFERENCE_v1` and its human provenance;
3. direct regime scope `A`, `T`, `RQ/R`, `S/K`;
4. four fixed wheels with the explicit second visit to `W_SK`;
5. exactly nine records in `BCDEFGHIK` order;
6. exact chamber codes `BBBBBB` through `KKKKKK`;
7. exact PRE and nine human-authored sentences;
8. perfect carrier alignment across every `S1-A-R-T-S2` record;
9. categorical `SER` only on the admitted diagonal records;
10. explicit ancestry for every declarative `R` value in the reference fixture.

## 3. Pair-space partition

The test derives rather than merely asserts the combinatorics:

```text
9 diagonal chambers: BB CC DD EE FF GG HH II KK
36 heterogeneous chambers: BC through IK
45 total unordered chambers
```

The two sets must be disjoint. Stack B owns the diagonal renderings. Stack G / Figura Tertia retains the heterogeneous medium-discovery space.

## 4. Adverse fixtures

| Case | Expected result | Required diagnostic |
|---|---|---|
| `AF_IN_E_CARRIER_MISALIGNMENT` | `REJECTED` | `CARRIER_MISALIGNMENT` |
| `ALIGNED_INSUFFICIENT_SUPPORT` | `SUSPENDED` | `PREDICATIVE_SUPPORT_INSUFFICIENT` |
| `ALIGNED_EXPLICIT_WEAKER_RELATION` | `CONTRACTED` | none |
| `RAW_Q_IN_R_SLOT` | `REJECTED` | `RQ_ANCESTRY_INCOMPLETE` |
| `V_RESULT_SCOPE_LEAK` | `REJECTED` | `V_RESULT_SCOPE_LEAK` |
| `EIGHT_POSITION_PADDING_REQUEST` | `REJECTED` | `INPUT_POSITIONAL_CARDINALITY` |
| `HETEROGENEOUS_MEDIUM_SHORTCUT` | `REJECTED` | `FIGURA_TERTIA_SCOPE_LEAK` |
| `AEGIS_RUNTIME_AUTHORITY` | `REJECTED` | `AEGIS_AUTHORITY_LEAK` |

## 5. Authority and containment checks

The validator also requires:

- all five artifacts remain explicitly `NON_RUNTIME` and are not marked canonical;
- no PDF, DOC, DOCX, or CARCER-named file exists inside the candidate;
- `SYSTEM_INDEX_CURRENT.md` does not register the candidate path;
- the separate A×T virtue/vice result family remains excluded from direct Stack B scope.

## 6. Deferred empirical tests

The static candidate does not yet close the deployment investigation. The active M2M must remain open until controlled low-token trials compare at least:

1. the four-wheel revisitation representation;
2. an equivalent six-ring expansion;
3. categorical, contracted, suspended, and rejected outputs;
4. context-bearing inputs with valid OPERA A contraction and OPERA S species evidence;
5. missing-provider and incomplete-RQ ancestry behavior;
6. round-trip preservation of the nine positional records.

Those trials may characterize behavior. They do not by themselves canonize the stack.

## 7. Non-deployment closure corpus

The low-token trials above are now owned by Backlog `M2M-DEPLOY-03` and are not a completion gate for the Stack B development slice.

`fixtures/STACK_B_NON_DEPLOYMENT_CLOSURE_FIXTURES.json` adds the in-scope closure cases: traceable synthetic provider interfaces, missing A/S/T/RQ evidence, independent single-role carrier swaps, ambiguity, contradiction, deliberate loss, overpredication, a contracted software-domain portability case, E→K and B→C structural proportional grids, canonical-to-applied reflection failure semantics, Stack G compatibility, and the `A_H`/derived-virtue boundary.

These fixtures test candidate contracts. They do not claim upstream OPERA execution or resolve the explicitly preserved semantic hypotheses.