{ "schema_version": "anvil-test-reduction-deleted-module-audit/v1", "base_revision": "7b5158b724ca715c85a3302091d2a2b40f253da0", "reviewer_role": "independent-adversarial-reviewer", "implementer_model_family": "GPT-6", "reviewer_model_family": "Sol", "reviewed_wholly_deleted_module_count": 62, "unreviewed_wholly_deleted_modules": [], "reviewed_modules": [ "tests/connect/test_isolation_declaration.py", "tests/connect/test_isolation_probe_regressions.py", "tests/connect/test_isolation_serial_pty.py", "tests/connect/test_qualification_config.py", "tests/connect/test_qualification_container_install.py", "tests/connect/test_session_lifetime.py", "tests/connect/test_user_backup_history.py", "tests/control_plane/test_controller_workload_cli.py", "tests/fixtures/operator_workflows/test_voice_latency_model_ab_matrix.py", "tests/media/test_comfyui_packaging.py", "tests/media/test_wan22_v2_workflow.py", "tests/media/test_workflow_compatibility.py", "tests/native_agents/test_adapter.py", "tests/observability/test_benchmark_session.py", "tests/observability/test_dashboard_workloads_ui.py", "tests/observability/test_history_compare.py", "tests/observability/test_milestone3_dashboard.py", "tests/observability/test_observatory_connect_access_ui.py", "tests/observability/test_workload_content_security_policy.py", "tests/observability/test_workload_review_regressions.py", "tests/router/test_bug_bash.py", "tests/router/test_dialect_parity.py", "tests/router/test_harness_contracts.py", "tests/router/test_image_fidelity.py", "tests/router/test_internal.py", "tests/router/test_reasoning_limit_probe.py", "tests/router/test_router_telemetry.py", "tests/router/test_video_fidelity.py", "tests/test_aider_surface.py", "tests/test_benchmark_context_evidence.py", "tests/test_benchmark_model_inventory.py", "tests/test_benchmark_request_controls.py", "tests/test_benchmark_terminal_punctuation.py", "tests/test_candidate_operations_skill.py", "tests/test_capacity_canary_prompt.py", "tests/test_capacity_concurrency.py", "tests/test_cli_json_error_contract.py", "tests/test_cli_reference_audit.py", "tests/test_docs_command_invocations.py", "tests/test_glm53_fp4_documentation.py", "tests/test_harness_runtime.py", "tests/test_hermes_catalog_snapshot.py", "tests/test_kernel_tuning_skill.py", "tests/test_markdown_links.py", "tests/test_model_pull_native_review.py", "tests/test_modularization_architecture.py", "tests/test_product_families.py", "tests/test_q36_manifest.py", "tests/test_release_sweep_evidence.py", "tests/test_stt_benchmark_skill.py", "tests/voice/test_llm_stage_incremental_emission.py", "tests/voice/test_mini_validation_defaults.py", "tests/voice/test_pipeline_spine.py", "tests/voice/test_tts_stage_incremental_emission.py", "tests/voice/test_tts_stream_config.py", "tests/voice/test_voice_evidence_publication.py", "tests/voice/test_voice_manifest_set.py", "tests/workbench/test_javascript_modules.py", "tests/workbench/test_packaging.py", "tests/workbench/test_pi_chat_behavior.py", "tests/workbench/test_pi_official_smoke.py", "tests/workbench/test_pi_web_mobile.py" ], "critical_restores": [ { "module": "tests/connect/test_isolation_declaration.py", "required_selectors": [ "tests/connect/test_isolation_declaration.py::test_static_declaration_matches_only_the_guest_declarative_functions" ], "reason": "Binds the static isolation deployment declaration to the closed declarative guest subset." }, { "module": "tests/connect/test_isolation_probe_regressions.py", "required_selectors": [ "tests/connect/test_isolation_probe_regressions.py::test_probe_against_real_unix_socket[reset-True-0]", "tests/connect/test_isolation_probe_regressions.py::test_probe_against_real_unix_socket[redirect-True-1]", "tests/connect/test_isolation_probe_regressions.py::test_probe_against_real_unix_socket[redirect-False-0]", "tests/connect/test_isolation_probe_regressions.py::test_missing_ingress_socket_is_not_a_passing_rejection", "tests/connect/test_isolation_probe_regressions.py::test_failure_injection_targets_the_declared_tunnel_listener" ], "reason": "Preserves fail-closed rejection semantics, real CRLF wire framing, missing-socket refusal, and exact failure-injection ownership." }, { "module": "tests/connect/test_isolation_serial_pty.py", "required_selectors": [ "tests/connect/test_isolation_serial_pty.py::test_guest_result_crosses_real_tty_as_one_canonical_frame" ], "reason": "Preserves the canonical serial result frame used by isolation qualification evidence." }, { "module": "tests/connect/test_qualification_config.py", "required_selectors": [ "tests/connect/test_qualification_config.py::test_unsafe_config_refuses_without_reading" ], "representative_rows": [ "symlink", "fifo", "oversized" ], "reason": "Configuration reads must not follow indirection, block on a FIFO, or accept unbounded input." }, { "module": "tests/connect/test_qualification_container_install.py", "required_selectors": [ "tests/connect/test_qualification_container_install.py::test_redirect_refuses_tls_downgrade_before_request", "tests/connect/test_qualification_container_install.py::test_only_verified_regular_binary_is_accepted" ], "representative_rows": [ "regular-valid", "symlink", "bad-hash" ], "reason": "Preserves TLS and archive-integrity boundaries before a qualification binary becomes executable." }, { "module": "tests/connect/test_session_lifetime.py", "required_selectors": [ "tests/connect/test_session_lifetime.py::test_browser_session_lifetime_is_optional_and_renders_exactly_when_declared", "tests/connect/test_session_lifetime.py::test_browser_session_lifetime_rejects_non_integer_or_out_of_range_values" ], "representative_rows": [ "0", "86401", "True" ], "reason": "Preserves browser-session expiry propagation and lower, upper, and bool-as-int rejection boundaries." }, { "module": "tests/control_plane/test_controller_workload_cli.py", "required_selectors": [ "tests/control_plane/test_controller_workload_cli.py::test_fleet_topology_does_not_enable_node_router_source" ], "reason": "A fleet topology must not infer or activate an undeclared router workload source." }, { "module": "tests/media/test_comfyui_packaging.py", "required_selectors": [ "tests/media/test_comfyui_packaging.py::test_public_runtime_scaffold_is_byte_identical_and_identity_free", "tests/media/test_comfyui_packaging.py::test_runtime_and_custom_node_inputs_are_exactly_pinned", "tests/media/test_comfyui_packaging.py::test_workflow_bundle_and_required_model_identities_are_pinned" ], "reason": "Preserves public scaffold synchronization, identity hygiene, and exact runtime, node, workflow, and model pins." }, { "module": "tests/media/test_workflow_compatibility.py", "required_selectors": [ "tests/media/test_workflow_compatibility.py::test_missing_node_is_normalized_without_private_metadata", "tests/media/test_workflow_compatibility.py::test_oversized_metadata_fails_closed" ], "reason": "Compatibility failures must be bounded and must not expose private backend metadata." }, { "module": "tests/native_agents/test_adapter.py", "required_selectors": [ "tests/native_agents/test_adapter.py::test_packaged_owner_adapter_behavior" ], "reason": "Executes the packaged owner adapter across bearer separation, one-shot interactions, drift, cancellation, generation fencing, and unsafe saved-path refusal." }, { "module": "tests/observability/test_observatory_connect_access_ui.py", "required_selectors": [ "tests/observability/test_observatory_connect_access_ui.py::test_connect_access_api_rejects_oversized_streams_and_malformed_tokens", "tests/observability/test_observatory_connect_access_ui.py::test_connect_access_view_disables_malformed_records_without_mutation", "tests/observability/test_observatory_connect_access_ui.py::test_connect_access_browser_boundary_is_same_origin_memory_only_and_fail_closed", "tests/observability/test_observatory_connect_access_ui.py::test_connect_access_view_uses_text_only_metadata_manual_refresh_and_no_account_creation" ], "reason": "Preserves browser-side auth, CSRF, same-origin, storage, malformed-data, bounds, and text-only privacy boundaries." }, { "module": "tests/observability/test_workload_content_security_policy.py", "required_selectors": [ "tests/observability/test_workload_content_security_policy.py::test_dashboard_document_and_packaged_workload_script_have_exact_safe_csp" ], "reason": "Preserves no-store, nosniff, same-origin and framing policy headers on the dashboard and packaged script." }, { "module": "tests/observability/test_workload_review_regressions.py", "required_selectors": [ "tests/observability/test_workload_review_regressions.py::test_updated_after_source_is_not_a_canonical_record[0:00:00.000001]", "tests/observability/test_workload_review_regressions.py::test_node_decoder_rejects_descendant_skew_against_node_collection", "tests/observability/test_workload_review_regressions.py::test_fleet_decoder_checks_every_descendant_against_fleet_collection[record]", "tests/observability/test_workload_review_regressions.py::test_fleet_source_time_failure_is_isolated_before_final_decode" ], "representative_rows": [ "updated-after-source:microsecond", "fleet-descendant:record" ], "reason": "Samples fail-closed temporal drift validation at record, node, fleet, and source-isolation seams." }, { "module": "tests/router/test_bug_bash.py", "required_selectors": [ "tests/router/test_bug_bash.py::test_decision_log_default_is_bounded", "tests/router/test_bug_bash.py::test_decision_log_rejects_nonpositive_cap" ], "reason": "Preserves the in-memory decision-log resource bound and invalid-cap refusal." }, { "module": "tests/router/test_harness_contracts.py", "required_selectors": [ "tests/router/test_harness_contracts.py::test_harness_protocols_retain_session_history_and_tool_continuations" ], "representative_rows": [ "pi-responses-X-Session-Affinity" ], "reason": "Preserves Pi affinity correlation into request history; no other retained test names this header." }, { "module": "tests/router/test_image_fidelity.py", "required_selectors": [ "tests/router/test_image_fidelity.py::test_same_dialect_openai_image_passthrough", "tests/router/test_image_fidelity.py::test_same_dialect_anthropic_image_passthrough" ], "reason": "Prevents silent image stripping while reconstructing relay request bodies for either supported dialect." }, { "module": "tests/router/test_router_telemetry.py", "required_selectors": [ "tests/router/test_router_telemetry.py::test_prometheus_is_gauge_only_and_escapes_safe_model_labels", "tests/router/test_router_telemetry.py::test_capacity_prometheus_rejects_malformed_replica_groups", "tests/router/test_router_telemetry.py::test_capacity_prometheus_enforces_per_tier_and_total_member_bounds" ], "reason": "Preserves label escaping, malformed-private-value suppression, and metrics-cardinality bounds." }, { "module": "tests/router/test_video_fidelity.py", "required_selectors": [ "tests/router/test_video_fidelity.py::test_same_dialect_openai_video_is_forwarded_verbatim", "tests/router/test_video_fidelity.py::test_cross_dialect_video_fails_closed_instead_of_flattening" ], "reason": "Prevents silent video stripping and unsupported cross-dialect flattening." }, { "module": "tests/test_aider_surface.py", "required_selectors": [ "tests/test_aider_surface.py::test_submit_preview_returns_frozen_profile_spec_without_store" ], "reason": "Dry-run preview must remain store-free while binding the frozen profile digest." }, { "module": "tests/test_harness_runtime.py", "required_selectors": [ "tests/test_harness_runtime.py::test_planner_and_adapter_reject_changed_plan_before_launch" ], "representative_rows": [ "aider", "performance" ], "reason": "Both harness adapters must reject plan drift before launching external work." }, { "module": "tests/test_hermes_catalog_snapshot.py", "required_selectors": [ "tests/test_hermes_catalog_snapshot.py::HermesCatalogSnapshotTests::test_failed_or_malformed_snapshot_never_falls_back_to_single_key_queries" ], "reason": "Malformed or failed native snapshots must not trigger hidden fallback queries." }, { "module": "tests/test_model_pull_native_review.py", "required_selectors": [ "tests/test_model_pull_native_review.py::test_different_repositories_share_disk_admission_lock", "tests/test_model_pull_native_review.py::test_receipt_creation_is_exclusive_and_never_replaces_another_file", "tests/test_model_pull_native_review.py::test_filtered_snapshot_refuses_prior_excluded_files_without_deletion", "tests/test_model_pull_native_review.py::test_receipt_repeated_writes_remain_complete_json", "tests/test_model_pull_native_review.py::test_downloader_version_rejects_python_interpreter", "tests/test_model_pull_native_review.py::test_real_process_records_bounded_hf_version" ], "reason": "Preserves acquisition admission, receipt custody, nondeletion, atomic completeness, and executable identity." }, { "module": "tests/test_q36_manifest.py", "required_selectors": [ "tests/test_q36_manifest.py::test_q36_launcher_validates_mtp_without_shell_evaluation", "tests/test_q36_manifest.py::test_q36_container_recipe_pins_source_images_model_and_gpu_boundary" ], "reason": "Preserves metacharacter-safe argument construction plus exact image, model, loopback, and GPU recipe boundaries." }, { "module": "tests/voice/test_llm_stage_incremental_emission.py", "required_selectors": [ "tests/voice/test_llm_stage_incremental_emission.py::test_first_sentence_reaches_output_queue_before_second_sentence_is_sent", "tests/voice/test_llm_stage_incremental_emission.py::test_barge_in_mid_stream_drops_subsequent_chunks_through_the_real_stage_thread" ], "reason": "Preserves real-thread incremental LLM delivery and cancellation of stale output." }, { "module": "tests/voice/test_tts_stage_incremental_emission.py", "required_selectors": [ "tests/voice/test_tts_stage_incremental_emission.py::test_first_audio_chunk_reaches_output_queue_before_second_chunk_is_sent" ], "reason": "Preserves real-thread incremental audio delivery rather than collect-then-return behavior." }, { "module": "tests/voice/test_tts_stream_config.py", "required_selectors": [ "tests/voice/test_tts_stream_config.py::test_manifest_stream_setting_reaches_real_pipeline_request", "tests/voice/test_tts_stream_config.py::test_stream_defaults_true_and_manifest_rejects_nonboolean", "tests/voice/test_tts_stream_config.py::test_declared_openai_voice_and_language_survive_live_pipeline_configuration", "tests/voice/test_tts_stream_config.py::test_explicit_generation_cap_has_live_benchmark_and_evidence_parity", "tests/voice/test_tts_stream_config.py::test_generation_cap_rejects_invalid_manifest_overlay_and_direct_config" ], "representative_rows": [ "stream:false", "max_tokens:none", "max_tokens:750", "invalid:true", "invalid:0", "invalid:8193" ], "reason": "Preserves stream, voice, language and bounded generation controls through manifest, live request, and evidence." }, { "module": "tests/voice/test_voice_evidence_publication.py", "required_selectors": [ "tests/voice/test_voice_evidence_publication.py::test_existing_evidence_is_preserved", "tests/voice/test_voice_evidence_publication.py::test_concurrent_evidence_publication_keeps_one_complete_attempt" ], "reason": "Evidence publication must be exclusive and atomic under existing-file and concurrent-writer races." }, { "module": "tests/voice/test_voice_manifest_set.py", "required_selectors": [ "tests/voice/test_voice_manifest_set.py::test_managed_voice_admission_loads_colocated_manifest_set" ], "reason": "Voice lifecycle admission must account for colocated manifest workloads sharing the GPU role." }, { "module": "tests/workbench/test_packaging.py", "required_selectors": [ "tests/workbench/test_packaging.py::test_pi_build_uses_packaged_pinned_sources_and_never_the_companion_checkout", "tests/workbench/test_packaging.py::test_runner_assets_are_byte_identical_to_reviewed_sources" ], "reason": "Preserves packaged-source authority and byte identity instead of using an ambient Companion checkout." }, { "module": "tests/workbench/test_pi_chat_behavior.py", "required_selectors": [ "tests/workbench/test_pi_chat_behavior.py::test_pi_chat_keeps_drafts_isolated_and_transcript_ordered" ], "reason": "Preserves per-session draft isolation, transcript order, and one-shot interaction control state." }, { "module": "tests/workbench/test_pi_web_mobile.py", "required_selectors": [ "tests/workbench/test_pi_web_mobile.py::test_packaged_mobile_api_behavior" ], "reason": "Executes packaged mobile auth, path and body bounds, idempotency, error privacy, native-session restoration, and event filtering." } ], "ordinary_modules": [ { "module": "tests/connect/test_user_backup_history.py", "classification": "retention sampling detail; critical digest refusal and restore nonactivation remain elsewhere" }, { "module": "tests/fixtures/operator_workflows/test_voice_latency_model_ab_matrix.py", "classification": "historical fixture field and prose assertions" }, { "module": "tests/media/test_wan22_v2_workflow.py", "classification": "exact graph literal and model sampling values" }, { "module": "tests/observability/test_benchmark_session.py", "classification": "capture history, stability, and hard-limit behavior retained in milestone and artifact tests" }, { "module": "tests/observability/test_dashboard_workloads_ui.py", "classification": "static UI structure and bounded-VM rendering" }, { "module": "tests/observability/test_history_compare.py", "classification": "dashboard history and comparison reporting detail" }, { "module": "tests/observability/test_milestone3_dashboard.py", "classification": "broad milestone integration and sampling cadence" }, { "module": "tests/router/test_dialect_parity.py", "classification": "structural reflection matrix duplicated by direct dialect behavior tests" }, { "module": "tests/router/test_internal.py", "classification": "direct-dispatch error prose and metadata shape" }, { "module": "tests/router/test_reasoning_limit_probe.py", "classification": "isolated research-probe fixture configuration" }, { "module": "tests/test_benchmark_context_evidence.py", "classification": "single benchmark narrative row" }, { "module": "tests/test_benchmark_model_inventory.py", "classification": "documentation inventory table schema" }, { "module": "tests/test_benchmark_request_controls.py", "classification": "request-control literals duplicated by retained preflight and remote CLI checks" }, { "module": "tests/test_benchmark_terminal_punctuation.py", "classification": "grader punctuation policy literals" }, { "module": "tests/test_candidate_operations_skill.py", "classification": "skill prose and discovery contract" }, { "module": "tests/test_capacity_canary_prompt.py", "classification": "benchmark prompt formatting and deterministic literal detail" }, { "module": "tests/test_capacity_concurrency.py", "classification": "benchmark reporting specificity for observed overlap" }, { "module": "tests/test_cli_json_error_contract.py", "classification": "duplicate CLI warning presentation detail" }, { "module": "tests/test_cli_reference_audit.py", "classification": "developer audit-script unit matrix; active audit still runs in CI" }, { "module": "tests/test_docs_command_invocations.py", "classification": "documentation command-shape audit" }, { "module": "tests/test_glm53_fp4_documentation.py", "classification": "documentation arithmetic" }, { "module": "tests/test_kernel_tuning_skill.py", "classification": "skill documentation and artifact naming checks" }, { "module": "tests/test_markdown_links.py", "classification": "developer link-checker parser matrix; link checker still runs in CI" }, { "module": "tests/test_modularization_architecture.py", "classification": "historical architecture audit; current documentation must not claim it remains a standing pytest gate" }, { "module": "tests/test_product_families.py", "classification": "product catalog and journey documentation shape" }, { "module": "tests/test_release_sweep_evidence.py", "classification": "historical release record consistency" }, { "module": "tests/test_stt_benchmark_skill.py", "classification": "skill and example artifact pin checks" }, { "module": "tests/voice/test_mini_validation_defaults.py", "classification": "default hostname pattern literals" }, { "module": "tests/voice/test_pipeline_spine.py", "classification": "echo-stage pipeline behavior duplicated by retained real-stage seam tests" }, { "module": "tests/workbench/test_javascript_modules.py", "classification": "broad syntax and run-coalescing UI matrix" }, { "module": "tests/workbench/test_pi_official_smoke.py", "classification": "optional external Pi smoke with environment-dependent skip" } ], "ordinary_coverage_within_critical_modules": "Any test or parameter row in a critical_restore module that is not listed above is classified as ordinary literal, duplicate, UI, format, or report specificity under the authorized noncritical-coverage tradeoff.", "limitations": [ "Every module wholly deleted in the reviewed snapshot was inspected; no wholly deleted module remains unreviewed.", "Partially reduced modules were reviewed through retained registries, safety selectors, high-risk removed-function searches, exact parameter-row comparison, and focused source inspection, but this does not prove semantic equivalence for every removed function or row across the full suite.", "Coverage percentage and test names were used only as discovery aids, not as proof that a behavior is preserved.", "Final acceptance still requires a fresh ordinary collection, all required selectors present, a clean full ordinary suite, combined statement-and-branch coverage at or above 70 percent, and an uninstrumented timed run for the runtime claim." ], "scope_note": "Sanitized independent whole-module deletion review. Final measured acceptance is recorded in test-suite-reduction-results.json; partial-file equivalence is not claimed." }