Knowledge Wiki Orchestrata Ultimate

Orchestrata Ultimate › Tests › Reference › gates

Tests reference — tests/gates

tests/gates — 8 test modules, 203 test functions. Source tree: tests/gates at commit 8a2cb42ad. Generated from the tests AST snapshot by scripts/gen-data-pages.py.

Prose pending.

Modules

Module Tests Covers
test_commit_review.py 34 2
test_differential_coverage.py 10 1
test_differential_vs_legacy.py 18 3
test_fact_extractor.py 44 3
test_freshness_inference.py 16 1
test_gate_explain.py 18 1
test_gate_rules.py 63 3

Module details

tests/gates/test_commit_review.py

source · 34 tests · 512 lines

Tests for the pre-commit review verdict derived in Prolog.

Covers: gates.commit_review, gates.engine

Kind Name Line Description
function test_review_required_mirrors_each_reviewers_own_scope 171
function test_a_name_matching_both_test_patterns_requires_one_review 177
function test_nothing_to_review_is_a_pass_that_says_so 182
function test_clean_reports_pass 192
function test_a_code_review_error_fails_and_names_the_finding 202
function test_code_review_below_error_does_not_block 213
function test_test_review_blocks_at_warning 218
function test_test_review_info_does_not_block 225
function test_a_repo_wide_finding_naming_another_file_still_blocks 230 code-review.py's banned-dependency and Prolog-boundary gates scan the
function test_a_required_review_that_never_ran_is_invalid 242
function test_a_reviewer_exit_that_is_not_a_verdict_is_invalid 250 2 is argparse rejecting a flag — the dead --format, exactly.
function test_a_verdict_exit_without_a_report_is_invalid 258 The json.load failure: the reviewer finished, but nothing parsed.
function test_a_report_over_zero_files_is_invalid 265
function test_a_stated_total_the_parsed_findings_do_not_match_is_invalid 272 The parser drifted from the report: it says 3, we read 1.
function test_an_unclassifiable_severity_is_invalid_once_per_severity 279
function test_an_exit_that_contradicts_the_report_is_invalid 295
function test_a_defective_review_has_no_pass_or_fail_answer 305 A count from a report that is not sound means nothing.
function test_a_review_the_rules_did_not_ask_for_is_invalid 312
function test_invalid_dominates_fail 319
function test_an_invalid_review_shows_the_reviewers_own_output 327
function test_a_fact_file_without_its_staged_count_is_invalid 348
function test_a_fact_file_missing_staged_paths_is_invalid 355 The driver stated 3 staged paths; only 1 reached the fact file.
function test_commit_status_has_exactly_one_answer 371
function test_runs_do_not_contaminate_each_other 379 janus keeps ONE SWI database per process. Run 1 asserts an ERROR; run 2,
function test_scope_does_not_leak_between_runs 388
function test_real_code_review_report_parses_and_adds_up 397
function test_real_test_review_report_parses_and_adds_up 409
function test_real_test_review_declines_a_conftest_with_no_report 422 Post-D10: handed a conftest EXPLICITLY, test-review.py analyzes it via T21
function test_staged_scripts_source_file_with_fabricated_env_key_routes_to_t21 435 W3 founding regression (ADR-0116): a staged scripts/*.py carrying a
function test_parsers_return_none_for_text_that_is_not_a_report 467
function test_end_to_end_bad_files_fail_with_every_finding 475
function test_end_to_end_clean_files_pass 485
function test_end_to_end_a_missing_reviewer_is_invalid 497 The checkout has no test reviewer: python exits 2 ("can't open file").
function test_end_to_end_a_reviewer_past_its_limit_is_invalid 507

tests/gates/test_differential_coverage.py

source · 10 tests · 247 lines

Coverage-completion tests for the pure comparator paths of the differential.

Covers: gates.differential

Kind Name Line Description
function test_finding_str_is_file_line_rule 38 The comparator's one-line identity: file:line [rule], never the message.
function test_difference_str_appends_detail_when_present 44
function test_difference_str_omits_suffix_without_detail 57
function test_render_reports_counts_then_differences_sorted 66 render() leads with the four counters, then one indented line per
function test_render_on_empty_report_has_no_detail_lines 115
function test_prolog_only_import_banned_in_registry_but_not_legacy_set 122 A registry entry the legacy hardcoded set lacks is a named GAIN:
function test_prolog_only_import_in_both_sets_is_unexplained 152 If the legacy rule ALSO bans the module, the registry-overlap mechanism
function test_prolog_only_known_key_in_legacy_list_is_unexplained 180 A key the legacy list already covers cannot explain the difference via
function test_prolog_only_no_module_no_key_no_keys_list_is_unexplained 204 Rule present, file inspected, but nothing at that line for either
function test_known_loss_causes_do_not_fail_equivalence 229 An OUT_OF_PROJECT_SCOPE legacy-only finding is one this layer is RIGHT

tests/gates/test_differential_vs_legacy.py

source · 18 tests · 494 lines

The Prolog gate against the checkers it replaces (P5).

Covers: gates.differential, gates.engine, gates.fact_extractor

Kind Name Line Description
function test_side_decides_direction_not_cause 489 is_loss asks which side found it, not why. A cause classifies the
class TestTheFourDefectsItFound 148 One test per defect the differential surfaced, so none can return.
method TestTheFourDefectsItFound.test_pyswip_is_banned_by_the_registry_not_by_a_literal 151 standards held _FORBIDDEN_IMPORTS = {"pyswip"} in code. The
method TestTheFourDefectsItFound.test_a_dotted_registry_entry_bans_the_submodule_only 165 unittest.mock is banned, unittest is not. A top-level-only
method TestTheFourDefectsItFound.test_a_project_root_of_dot_does_not_silence_every_rule 180 project(_, '.') made in_project/1 fail for every file, and
method TestTheFourDefectsItFound.test_the_module_lookup_is_asked_only_about_import_rules 196 A relative-import finding at a line importing .relative was
class TestLosses 219
method TestLosses.test_a_package_only_the_legacy_checker_bans_is_a_loss 220
method TestLosses.test_a_classified_loss_still_fails 240 Knowing WHY a hole exists does not fill it. A cause is an
method TestLosses.test_a_substring_only_match_is_not_a_loss 261 _check_no_mocks flags any module whose name CONTAINS "mock", so
class TestGains 285
method TestGains.test_a_rule_no_legacy_checker_has_is_a_named_gain 286
method TestGains.test_a_file_the_legacy_rule_never_inspected_is_a_named_gain 300 _check_no_mocks only reads pytest files; banned_import reads
method TestGains.test_an_infra_key_the_legacy_list_omits_is_a_named_gain 321 ENV_VAR_KEYS covers 6 of the 10 required infra keys.
method TestGains.test_a_gain_never_fails_the_differential 342 Equality of the two sets is the WRONG acceptance test: it would
class TestScope 359
method TestScope.test_a_legacy_finding_outside_the_project_is_not_a_loss 360
class TestBanTargets 377
method TestBanTargets.test_both_granularities_are_offered 386 Mirrors ban_target/3: the comparison does not know whether the
class TestMessages 395
method TestMessages.test_a_relative_import_with_no_module_names_its_dots 396 from . import sibling has no module, and interpolating the empty
class TestRegistryAgreement 413
method TestRegistryAgreement.test_the_prolog_rule_flags_every_package_in_the_registry 414 The registry is read in Python (banned_registry) and matched in
class TestCLIRefusal 440 The ledger is empty. Both refusals below are the LIVE behaviour.
method TestCLIRefusal.test_the_cli_refuses_when_no_legacy_checker_is_left 443 Every row is deleted, so a comparison would report zero differences
method TestCLIRefusal.test_a_refilled_map_with_no_collector_fails_loudly 462 P7 refills the map for the next slice. If it refills the map and

tests/gates/test_fact_extractor.py

source · 44 tests · 989 lines

Tests for the Prolog gate fact extractor (P1).

Covers: core.infra_env_rule, core.janus_bootstrap, gates.fact_extractor

Kind Name Line Description
class TestFactsComeFromTheSnapshot 108 REQ-GATE-001 — the extractor must not open a source file.
method TestFactsComeFromTheSnapshot.test_structural_facts_ignore_the_on_disk_bytes 111 The snapshot says the module imports json; disk says otherwise.
method TestFactsComeFromTheSnapshot.test_a_file_absent_from_disk_still_yields_its_structural_facts 152 Nothing is read, so nothing needs to be there.
method TestFactsComeFromTheSnapshot.test_a_missing_snapshot_raises_instead_of_yielding_zero_facts 168 Zero facts means zero violations means a green gate (REQ-GATE-003).
method TestFactsComeFromTheSnapshot.test_a_snapshot_that_is_not_an_object_raises 181
class TestFreshnessIsTheOnlyDiskAccess 189 REQ-GATE-002 — staleness needs both sides, so it may touch disk.
method TestFreshnessIsTheOnlyDiskAccess.test_both_hashes_are_emitted_when_the_file_exists 192
method TestFreshnessIsTheOnlyDiskAccess.test_a_snapshot_file_gone_from_disk_emits_no_disk_side 221 The MISSING disk fact is what freshness.pl reads as staleness.
method TestFreshnessIsTheOnlyDiskAccess.test_snapshot_built_at_is_the_snapshots_own_mtime 250
class TestEnvAccessFacts 259 The two slice-1 env rules read this predicate and disagree about it.
method TestEnvAccessFacts.test_form_folds_the_default_in 276
method TestEnvAccessFacts.test_env_access_reaches_the_fact_base_with_its_line_and_key 281
class TestArgvLiteralFacts 309 ADR-0089 (Q2b): the spawn-site program token reaches the fact base.
method TestArgvLiteralFacts.test_argv_literal_is_emitted_with_its_exact_arguments 318
method TestArgvLiteralFacts.test_the_program_token_is_emitted_verbatim 348 An absolute system path is a record like any other — WHICH tokens
method TestArgvLiteralFacts.test_an_unrenderable_callee_keeps_its_question_mark 376 The analyzer records '?' when the callee chain is not
class TestSnapshotCaptureFacts 405 Provenance: which per-module captures the snapshot attests it performed.
method TestSnapshotCaptureFacts.test_one_fact_per_capture_entry 415
method TestSnapshotCaptureFacts.test_no_captures_key_emits_no_capture_facts 431 The pre-Q2a snapshot shape. Emitting a fact for a capture nobody
class TestPolicyData 444
method TestPolicyData.test_infra_keys_are_emitted_from_the_injected_set 445
method TestPolicyData.test_the_live_infra_key_set_is_the_default 459 The gate must not carry its own copy of the infra-key list.
method TestPolicyData.test_banned_imports_carry_their_rationale 474
method TestPolicyData.test_a_version_pin_is_stripped_to_the_module_name 496
class TestDifferentialLookups 510 The two lookups the P5 differential asks the extractor for.
method TestDifferentialLookups.test_the_import_index_keeps_the_full_dotted_path 518 Not folded to the top-level package.
method TestDifferentialLookups.test_the_env_index_carries_the_key_read_at_each_line 538 The legacy ENV_VAR_KEYS list covers 6 of the 10 required infra
method TestDifferentialLookups.test_the_import_index_excludes_files_outside_the_project 561 A legacy-only finding in an out-of-scope file is a scope difference,
method TestDifferentialLookups.test_the_registry_is_normalised_for_comparison 581 PyPI treats - and _ as the same character, and so does
method TestDifferentialLookups.test_file_class 612
class TestPrologSyntax 616
method TestPrologSyntax.test_atoms_are_escaped 629 One unescaped newline in one default expression would fail the
method TestPrologSyntax.test_empty_list_renders_as_prolog_nil 634
method TestPrologSyntax.test_list_elements_are_atoms 637
method TestPrologSyntax.test_a_default_expression_with_a_newline_survives 640
class TestFactBaseLayout 661
method TestFactBaseLayout.test_facts_are_grouped_by_predicate 662 Interleaved clauses make SWI warn once per run boundary — measured
method TestFactBaseLayout.test_every_emitted_predicate_is_declared_in_sections 696 A predicate the layout does not know about lands under an explicit
method TestFactBaseLayout.test_write_persists_the_fact_base 708
class TestToolingLayer 718 REQ-GATE-010/011 — the linter as a fact source, not a thing to reimplement.
method TestToolingLayer.test_the_live_catalogue_and_its_configured_status_are_facts 727
method TestToolingLayer.test_e501_is_recorded_as_ignored_which_is_what_makes_it_derivable 743 The single fact pair P3's uncovered/1 turns into a conclusion.
method TestToolingLayer.test_a_catalogue_that_cannot_be_obtained_is_NAMED 773 Zero catalogue facts must never be silent.
method TestToolingLayer.test_a_catalogue_that_hangs_is_reported_as_timeout 795
method TestToolingLayer.test_rule_status_keeps_ignored_and_unknown_apart 820 A deliberate exemption and an oversight must not look the same —
method TestToolingLayer.test_fixability_survives_either_ruff_field 837
method TestToolingLayer.test_a_legacy_top_level_ruff_config_is_still_read 842 A config that moved must not silently report every rule unknown —
method TestToolingLayer.test_a_missing_pyproject_yields_no_selectors 857
class TestConsultsIntoARealEngine 869 The emitted text must be valid Prolog, and nothing else proves it.
method TestConsultsIntoARealEngine.test_adversarial_atoms_consult_without_error 879
method TestConsultsIntoARealEngine.test_the_live_fact_base_consults_and_answers 905 88k facts through the real engine, then a slice-1-shaped query.
class TestLiveTreeAcceptance 937 P1's acceptance lines, asserted rather than eyeballed.
method TestLiveTreeAcceptance.test_the_ontology_declares_no_rule_shaped_predicate 940 REQ-GATE-004. A clause with a body in code_facts.pl means policy
method TestLiveTreeAcceptance.test_distinct_fact_predicates_stay_within_the_ceiling 954
method TestLiveTreeAcceptance.test_no_per_line_fact_layer_exists 961 REQ-GATE-010. A per-line layer was measured at 845,547 facts for the
method TestLiveTreeAcceptance.test_the_live_tree_yields_facts_for_every_module 983

tests/gates/test_freshness_inference.py

source · 16 tests · 423 lines

Tests for freshness-as-inference and the three-valued verdict (P2).

Covers: gates.engine

Kind Name Line Description
class TestAStaleFactBaseIsInvalidNotPass 189 REQ-GATE-003, the acceptance line: touching a file must not exit 0.
method TestAStaleFactBaseIsInvalidNotPass.test_a_touch_that_changes_no_bytes_is_NOT_stale 192 Mohamed, 2026-08-08: "compare the file hash not the time."
method TestAStaleFactBaseIsInvalidNotPass.test_a_content_change_with_an_UNCHANGED_mtime_is_still_caught 209 The case an mtime-only check misses entirely.
method TestAStaleFactBaseIsInvalidNotPass.test_a_file_the_snapshot_describes_but_disk_lost_is_caught 231
method TestAStaleFactBaseIsInvalidNotPass.test_a_file_on_disk_the_snapshot_never_saw_is_caught 242 The worst blind spot of the four.
method TestAStaleFactBaseIsInvalidNotPass.test_staleness_is_reported_before_any_rule_runs 263 A stale base yields NO violations, even under an always-firing rule.
class TestNoRulesIsAlsoInvalid 279
method TestNoRulesIsAlsoInvalid.test_zero_rules_does_not_report_a_pass 280 Zero violations found by zero rules is not a pass.
method TestNoRulesIsAlsoInvalid.test_the_live_tree_with_no_rulebase_is_invalid 290 The state of this repo at P2, asserted rather than assumed.
class TestAFreshBaseWithRulesReachesAVerdict 306
method TestAFreshBaseWithRulesReachesAVerdict.test_rules_that_find_nothing_produce_pass 307
method TestAFreshBaseWithRulesReachesAVerdict.test_rules_that_find_something_produce_fail 317
method TestAFreshBaseWithRulesReachesAVerdict.test_the_three_exit_codes_are_distinct 330 A caller that only checks rc == 0 must still be stopped by
class TestRunsDoNotContaminateEachOther 337 janus shares ONE process-global SWI database.
method TestRunsDoNotContaminateEachOther.test_a_second_run_does_not_inherit_the_first_runs_facts 340 The exact leak measured while building P2.
method TestRunsDoNotContaminateEachOther.test_a_run_that_raises_still_unloads 369 A syntactically broken rulebase must not strand 90k facts.
class TestVerdictReporting 394
method TestVerdictReporting.test_truncation_is_announced 395 A list silently cut at N reads as 'there were N' — which is how a
method TestVerdictReporting.test_a_short_list_is_not_annotated 408
method TestVerdictReporting.test_a_violation_renders_as_a_clickable_location 413
method TestVerdictReporting.test_exit_code_follows_status 422

tests/gates/test_gate_explain.py

source · 18 tests · 319 lines

Proof trees and why-not, over the real meta-interpreter (P4).

Covers: gates.engine

Kind Name Line Description
class TestExplain 118
method TestExplain.test_a_violation_yields_a_proof_naming_the_facts_it_consumed 119 The point of a proof: every step is inspectable, so a wrong finding
method TestExplain.test_the_proof_distinguishes_facts_from_inference 134 A uniform tree hides the only question a reader has: which steps
method TestExplain.test_the_rendered_tree_marks_each_kind 144
method TestExplain.test_the_project_scope_check_appears_in_the_proof 155 in_project/1 is not decoration — a proof that omitted it would
method TestExplain.test_the_bare_toolchain_proof_names_the_fact_and_the_enumeration 170 ADR-0089 (Q2b): the proof of a bare_toolchain finding must show
method TestExplain.test_a_location_that_does_not_violate_yields_no_proof 188 None, not an empty tree. An empty proof renders as 'explained'.
method TestExplain.test_an_unknown_rule_id_yields_no_proof 198
class TestWhyNot 206 REQ-GATE-008 — the question the tree-sitter checker died for want of.
method TestWhyNot.test_an_absolute_import_names_the_level_comparison_WITH_ITS_VALUE 209 0>0, not _G>0.
method TestWhyNot.test_a_correct_infra_read_names_the_form_that_saved_it 226 REDIS_HOST read WITHOUT a default is the required form, and the
method TestWhyNot.test_an_ordinary_key_with_a_default_names_the_infra_key_check 239
method TestWhyNot.test_a_non_banned_import_names_the_banned_lookup 249
method TestWhyNot.test_a_line_with_no_import_at_all_names_the_missing_fact 257 Different cause, different answer: there is no import HERE, as
method TestWhyNot.test_an_unknown_rule_has_no_candidate_clause 269 Empty, and the CLI turns that into a message rather than silence —
method TestWhyNot.test_every_rule_can_be_asked_why_not 288 A rule that cannot answer why-not is a rule that can die silently.
class TestRenderers 300
method TestRenderers.test_a_leaf_renders_without_children 301
method TestRenderers.test_nesting_is_indented 304
method TestRenderers.test_an_unknown_kind_renders_a_question_mark 311 A node kind explain.pl gains later must be VISIBLE, not silently
method TestRenderers.test_a_why_not_names_both_the_clause_and_the_blocker 316

tests/gates/test_gate_rules.py

source · 63 tests · 1379 lines

Slice-1 rules, with a negative corpus at least as large as the positive (P3).

Covers: core.janus_bootstrap, gates.engine, gates.fact_extractor

Kind Name Line Description
class TestNoRelativeImport 305
method TestNoRelativeImport.test_positive 309
method TestNoRelativeImport.test_negative 318
method TestNoRelativeImport.test_the_detail_names_the_depth_and_the_module 323 A finding that does not say WHAT is wrong makes the reader re-derive
class TestBannedImport 335
method TestBannedImport.test_positive 339
method TestBannedImport.test_negative 347
method TestBannedImport.test_the_detail_carries_the_ban_reason 352 "xdist is banned" is a wall. The registry records WHY, and carrying
class TestEnvMutation 363 The write, split from the read on 2026-08-08.
method TestEnvMutation.test_positive 375
method TestEnvMutation.test_negative 383
method TestEnvMutation.test_a_write_does_not_also_fire_the_read_rule 388 The defect itself: one access, one rule. Two findings on one line
method TestEnvMutation.test_a_read_does_not_fire_the_mutation_rule 395
method TestEnvMutation.test_an_infra_key_read_bare_fires_nothing 400 The two env rules must not contradict each other on one line.
class TestBareOsEnviron 413
method TestBareOsEnviron.test_positive 417
method TestBareOsEnviron.test_negative 425
class TestWaivers 431 The repo's # gate-ok: <reason> convention, honoured by the rules.
method TestWaivers.test_a_reasoned_waiver_clears_a_mutation 442
method TestWaivers.test_the_same_mutation_fires_without_one 449 The control. Without it the test above would pass on a rule that
method TestWaivers.test_a_reasoned_waiver_clears_a_bare_read 456
method TestWaivers.test_a_bare_marker_with_no_reason_waives_nothing 461 An unexplained waiver is a mute button. Every checker that honours
method TestWaivers.test_a_marker_on_the_line_above_reaches_the_finding 473 The form the repo uses when the statement cannot carry a trailing
method TestWaivers.test_a_marker_two_lines_above_does_not_reach 484 Bounded on purpose — a waiver that drifted down a file would cover
method TestWaivers.test_a_waiver_does_not_reach_banned_import 493 gate_severity.pl says banned_import has NO exception in any context.
class TestInfraKeyWithDefault 526
method TestInfraKeyWithDefault.test_positive 530
method TestInfraKeyWithDefault.test_negative 538
method TestInfraKeyWithDefault.test_it_is_the_exact_inverse_of_bare_os_environ 543 The two rules over env_access/4 disagree on purpose, and the pair is
class TestBareToolchain 558 ADR-0089 (Q2b): a literal toolchain program token at a spawn site.
method TestBareToolchain.test_positive 579 Exactly ONE violation per record, with the enumerated Kind in the
method TestBareToolchain.test_negative 595
method TestBareToolchain.test_the_detail_carries_the_remedy 600 "bare toolchain name python" is a wall. The remedy is DATA
method TestBareToolchain.test_a_python_tool_remedy_names_the_module_form 612
method TestBareToolchain.test_a_waived_line_does_not_fire 620 The 4 test-data records in this tree (a program name passed to a
method TestBareToolchain.test_the_same_record_fires_without_the_waiver 628 The control. Without it the test above would pass on a rule that
method TestBareToolchain.test_a_record_outside_the_project_scope_does_not_fire 635 in_project/1 is defence in depth behind the extractor's scope
class TestRequiredCaptures 665 A rule whose input was never captured matches nothing and reports PASS.
method TestRequiredCaptures.test_a_snapshot_without_argv_literals_is_invalid 676
method TestRequiredCaptures.test_a_snapshot_without_env_accesses_is_invalid 685
method TestRequiredCaptures.test_a_snapshot_with_no_capture_list_at_all_is_invalid 694 The pre-Q2a snapshot shape: no captures key means NO capture is
method TestRequiredCaptures.test_both_captures_present_with_zero_records_passes 708 Captured and found nothing IS clean — the distinction the capture
class TestSeverityIsSeparateData 718 REQ-GATE-006 — enforcement policy lives outside the rule clauses.
method TestSeverityIsSeparateData.test_no_rule_clause_mentions_a_severity 721 The check that keeps the two files honest.
method TestSeverityIsSeparateData.test_no_severity_clause_mentions_a_fact_predicate 735
method TestSeverityIsSeparateData.test_a_src_violation_blocks 745
method TestSeverityIsSeparateData.test_an_infra_key_default_blocks_even_in_generated_code 754 gate_severity.pl grants NO exception for this rule, anywhere.
class TestProjectScope 772 Mohamed, 2026-08-08: "make sure to scope facts by project".
method TestProjectScope.test_a_rule_does_not_fire_outside_the_declared_project 775 in_project/1 is defence in depth behind the extractor's scope
method TestProjectScope.test_a_fact_base_with_no_declared_scope_is_invalid_not_clean 808 The false-PASS this whole layer exists to refuse.
method TestProjectScope.test_the_project_fact_names_the_snapshot 848
class TestCorpusDiscipline 854 The property all four 2026-08-06 hooks were missing.
method TestCorpusDiscipline.test_negative_corpus_is_at_least_as_large_as_positive 884
method TestCorpusDiscipline.test_every_declared_rule_has_a_corpus 893 A rule in gate_rules.pl with no test here would be the tree-sitter
class TestPolicyMechanismFacts 914 REQ-GATE-011: policy_mechanism/3, emitted from what is ACTUALLY in the
method TestPolicyMechanismFacts.test_every_native_rule_gets_its_own_mechanism_fact 922 One policy_mechanism(RuleId, native, RuleId) per rule
method TestPolicyMechanismFacts.test_a_missing_gate_rules_file_emits_no_native_mechanisms 950 The injection seam works, and fails OPEN toward "nothing claimed"
method TestPolicyMechanismFacts.test_ruff_mechanisms_mirror_the_catalogue_exactly 969 Every ruff-side policy_mechanism fact has a matching tool_rule/4
class TestFormatterMechanism 1010 Coordinator finding, 2026-08-19: covered/1 must not be blind to a
method TestFormatterMechanism.test_black_configured_emits_a_formatter_mechanism_for_trailing_whitespace 1023
method TestFormatterMechanism.test_black_configured_emits_a_formatter_mechanism_for_all_four_codes 1035 W191/W291/W292/W293 — every code
method TestFormatterMechanism.test_e501_never_gets_a_formatter_mechanism 1060 The regression pin for the false positive the coordinator found:
method TestFormatterMechanism.test_no_black_section_emits_no_formatter_mechanisms 1081 The condition IS the fact: no [tool.black], no formatter claim.
class TestPolicyCoverage 1099 REQ-GATE-011: covered/1, uncovered/1, policy_requirement/1 — RUL-014
method TestPolicyCoverage.test_a_native_policy_with_a_live_rule_is_covered_not_uncovered 1127 gate_rules.pl's OWN violation/4 clause backs this policy — it must
method TestPolicyCoverage.test_an_ignored_ruff_policy_is_uncovered 1143 A policy whose ONLY mechanism is a ruff rule this project's config
method TestPolicyCoverage.test_an_enabled_ruff_policy_is_covered_not_uncovered 1158 This is the "IS covered" half again, over the TOOL mechanism this
method TestPolicyCoverage.test_a_stale_native_mechanism_is_uncovered_not_covered 1175 native_rule_live/1 re-checks at QUERY time: a policy_mechanism
method TestPolicyCoverage.test_unknown_ruff_status_is_not_a_policy_requirement 1190 A ruff rule this project has never selected is not one of ITS
method TestPolicyCoverage.test_a_formatter_mechanism_is_covered_not_uncovered 1213 The THIRD mechanism kind (coordinator finding, 2026-08-19):
method TestPolicyCoverage.test_a_formatter_mechanism_is_also_a_policy_requirement 1229 A formatter-only-mechanised policy must be able to be ASKED
class TestClauseCountInvariant 1247 REQ-114-FR2-004 clause (a) (SPEC-114 FR-002): the violation/4 clause-count
method TestClauseCountInvariant.test_the_real_gate_rules_pl_loads_every_declared_clause_and_passes 1265 The clean-tree control: no clobbering, 6 declared, 6 loaded,
method TestClauseCountInvariant.test_a_second_moduleless_rulebase_defining_violation4_is_invalid 1276 THE NEGATIVE CORPUS. A quarantine-shaped pending.pl — no
method TestClauseCountInvariant.test_the_clobber_reason_names_the_survivor_and_the_wiped_file 1308 The verdict must say WHICH file clobbered it, not just that the
method TestClauseCountInvariant.test_consult_order_decides_who_survives_and_the_message_tracks_it 1332 Consulting pending.pl FIRST reverses which file wins — SWI
method TestClauseCountInvariant.test_no_rulebases_still_falls_through_to_the_zero_rules_reason 1354 0 declared == 0 loaded satisfies the invariant trivially, so the
method TestClauseCountInvariant.test_declared_rule_count_of_a_missing_file_is_zero 1372 A rulebases entry that does not exist on disk counts as zero

Support files

Fixtures, package markers and helpers without test functions.