fix(ontology): move :AnnexIII_Condition_1a and :AnnexIII_Condition_5b to governance extension

[Amosk21]

Why

Deduplicates universal regulatory content out of two per-fixture instance files into ARCO_governance_extension.ttl. Closes regulatory_alignment FAIL and traceability FAIL on DecoySystem_001 and GhostSystem_001 by making the regulatory condition declarations visible to every fixture that imports the governance extension. Extended scope (2026-05-14) refreshes stale doc/header references and adds two coverage gaps the README previously disclosed.

Both audit queries (check_regulatory_alignment.sparql, check_assessment_traceability.sparql) require the condition to be typed :RegulatoryContent in the merged graph. The conditions were previously declared only inside ARCO_instances_sentinel.ttl (1(a)) and ARCO_instances_creditscoring.ttl (5(b)). Fixtures that don't import either instance file referenced the conditions via iao:0000136 but the type assertion wasn't present, so both audit queries returned FAIL for fixture-distribution reasons, not fixture-semantics reasons.

What changes

Three commits.

Commit 1: ontology migration (390e115)

• 03_TECHNICAL_CORE/ontology/ARCO_governance_extension.ttl: new section "3a) REGULATORY CONTENT" adds :AnnexIII_List, :AnnexIII_Condition_1a, :AnnexIII_Condition_5b with the same triples previously in the two instance files. Triples preserved verbatim.
• 03_TECHNICAL_CORE/ontology/ARCO_instances_sentinel.ttl: removes 10 lines (declaration block); section header replaced with a brief migration comment. References preserved.
• 03_TECHNICAL_CORE/ontology/ARCO_instances_creditscoring.ttl: removes 13 lines (declaration block + self-containedness comment); section header replaced with a brief migration comment. References preserved.

Commit 2: stale-doc refresh (cffcf2e)

• ARCO_instances_flag_tests.ttl header: replaces pre-migration text ("audit FAIL... minimal instances for flag testing... without full regulatory content linkage") with the actual post-migration state. Names the remaining cause of traceability FAIL and regulatory_alignment FAIL (local :AssessmentDocumentation -> :AnnexIII_Condition_* iao:0000136 link absent from this fixture, not fixture-distribution).
• LIMITATIONS.md sec 9: file reference for the class-as-individual cco:prescribes triple updated from ARCO_instances_sentinel.ttl to ARCO_governance_extension.ttl. Adds the 5(b) companion triple. Notes the new 5(b) gate-removal coverage and adversarial-mechanism tests.
• README.md: "Gate independence is empirically verified" sentence drops the "(Symmetric coverage for 5(b) is queued.)" parenthetical; corresponding row in the active-changes table moves from "Active work" to "Landed 2026-05-14".

Commit 3: coverage additions (d74e76a)

• test_gate_removal.py: parameterized over both modeled Annex III categories. Original 7 Sentinel tests (5 removals + 2 mutations) preserved by triple; 7 symmetric CreditScorer tests added against AnnexIII5bApplicableSystem. Closes the previously disclosed 5(b) gap.
• test_adversarial_mechanism.py (new): asserts that DecoySystem_001 Gate 1 routes via owl:equivalentClass propagation (:WeirdScanner typed pre-reasoning, :BiometricIdentificationCapability absent from asserted triples and entailed post-reasoning) and that GhostSystem_001 disposition is a blank node still satisfying owl:someValuesFrom. test_scenarios.py asserts the entailment fires; this test asserts HOW.
• .github/workflows/arco-smoke-test.yml and arco-demo.yml: both workflows run the new test alongside the existing three regression tests.

Tests

Per-fixture diff table (pre vs post commit 1, all 7 fixtures, 5 output channels):

Fixture	summary.json field deltas (semantic)	SHACL	Classification
Sentinel_ID_System	entailed_triples_added 20160 -> 20194 only	PASS unchanged	unchanged
CreditScorer_001	entailed_triples_added 20103 -> 20137 only	PASS unchanged	unchanged
VerificationKiosk_001	entailed_triples_added 20025 -> 20114 only	PASS unchanged	unchanged
DecoySystem_001	traceability FAIL->PASS; regulatory_alignment FAIL->PASS; triples +89	PASS unchanged	unchanged
GhostSystem_001	traceability FAIL->PASS; regulatory_alignment FAIL->PASS; all_checks_passed false->true; triples +89	PASS unchanged	unchanged
FlagTest_BiometricSystem_WithDerogationClaim	entailed_triples_added +89 only	PASS unchanged	unchanged
FlagTest_CreditSystem_WithFraudProcess	entailed_triples_added +89 only	PASS unchanged	unchanged

Regression suite (all PASS post commit 3):

• test_gate_removal.py (now covers both 1(a) and 5(b))
• test_scenarios.py (7 scenarios, including FlagTest expected_flags)
• test_kiosk_html_no_false_concretization.py
• test_adversarial_mechanism.py (new — 2 mechanism assertions for Decoy and Ghost)
• test_output_provenance.py 1 failure (unchanged baseline)
• hermit_cross_check.py: HermiT and OWL-RL agree on every (fixture, system, query) in the certificate-grade set

Coverage audit summary (added 2026-05-14):

ARCO claim	Test that proves it	Coverage
Two-reasoner cross-check (OWL-RL + HermiT agree)	hermit_cross_check.py	covered
Each Annex III 1(a) gate independently necessary	test_gate_removal.py CATEGORY_1A	covered
Each Annex III 5(b) gate independently necessary	test_gate_removal.py CATEGORY_5B	covered (new)
Negative case (verification kiosk) does NOT entail 1(a)	test_scenarios.py	covered
Cross-category isolation	test_scenarios.py	covered
Decoy: OWL inference fires via owl:equivalentClass	test_adversarial_mechanism.py	covered (new)
Ghost: OWL inference fires via blank-node disposition	test_adversarial_mechanism.py	covered (new)
FlagTest classification + flag fire simultaneously	test_scenarios.py expected_flags	covered
HTML output discipline (kiosk)	test_kiosk_html_no_false_concretization.py	covered
Output provenance contract	test_output_provenance.py	failing-by-design, 1 baseline

Downstream consumer audit (grep across 03_TECHNICAL_CORE/, docs/, mcp/, .github/): all reference sites either load ARCO_governance_extension.ttl (via every pipeline / test / cross-check loader confirmed at run_pipeline.py:1902, test_scenarios.py:153, hermit_cross_check.py:83, test_gate_removal.py:204) or are documentation mentions of the IRI itself. No consumer depends on the conditions being declared in a specific instance file.

Deviation from plan

The migration plan predicted four fixtures would flip regulatory_alignment FAIL->PASS: Decoy, Ghost, FlagTest_BiometricSystem_WithDerogationClaim, FlagTest_CreditSystem_WithFraudProcess. Actual outcome: only Decoy and Ghost flip. The two FlagTest fixtures do not flip because their :AssessmentDocumentation instances (ARCO_instances_flag_tests.ttl:90-92, 155-157) do not contain any iao:0000136 :AnnexIII_Condition_* triple; the audit query's ?doc iao:0000136 ?condition path is empty independent of where the condition is declared. This is a separate fixture-authoring gap, not introduced by this PR. The FlagTest fixture header (commit 2) now documents this state explicitly.

Deferred

• :AnnexIII_Condition_1a_Exclusion in ARCO_instances_verification.ttl:25-29 is a different class (verification-kiosk exclusion documentation per Recital 22 / Art 3(41)). Whether to also generalize the exclusion pattern is a separate future decision.
• FlagTest fixtures' :AssessmentDocumentation instances do not link to any regulatory condition; closing their regulatory_alignment FAIL is a separate fixture-authoring change.
• Negative-case isolated pipeline harness (LIMITATIONS sec 9 remains queued; not closed by this PR).

Revert

git revert <commit-hash> cleanly undoes each commit in isolation. Commits 2 and 3 are documentation and test-only and have no production-behavior impact. Reverting commit 1 sends the audit query back to FAIL on Decoy and Ghost (pre-existing state on main) with no other downstream consumer impact because the merged ontology is structurally identical.

[greptile-apps]

Your free trial has ended. If you'd like to continue receiving code reviews, you can add a payment method here.

[greptile-apps]

Your free trial has ended. If you'd like to continue receiving code reviews, you can add a payment method here.