Upload 58 files

tests/test_v52_ui_fixes.py

@@ -35,6 +35,6 @@ def test_runs_explorer_is_limited_to_five_rows_with_pagination():
 
 def test_progress_endpoint_uses_view_model_links_for_job_url():
     app_py = Path("app.py").read_text(encoding="utf-8")
-    assert '"version": "v81-robust-timeline-color-feedback"' in app_py
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in app_py
     assert 'view = build_run_view_model(run_id, bundle, bucket_source=bucket_source)' in app_py
     assert '"links": view.get("links")' in app_py

tests/test_v57_live_pi_traces.py

@@ -32,4 +32,4 @@ def test_ui_has_pi_live_trace_panel_and_renderer():
 
 
 def test_app_version_v57():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v58_stable_refresh_pi_stdout.py

@@ -23,4 +23,4 @@ def test_elapsed_ticks_client_side_each_second():
 
 
 def test_app_version_v62():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v59_job_log_progress_clean.py

@@ -51,4 +51,4 @@ def test_progress_renderer_no_longer_sets_active_run_link_buttons():
 
 
 def test_app_version_v59():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v60_cancel_persist_delete_run.py

@@ -44,4 +44,4 @@ def test_active_run_still_only_has_cancel_button():
 
 
 def test_app_version_v60():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v61_stabilize_buttons_runtime.py

@@ -45,4 +45,4 @@ setTimeout(()=>process.exit(0), 25);
 
 
 def test_app_version_v61():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v62_progress_perf_clean_delete.py

@@ -40,4 +40,4 @@ def test_delete_selected_run_clears_ui_and_refreshes_runs():
 
 
 def test_app_version_v62():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v63_delete_import_fix.py

@@ -14,4 +14,4 @@ def test_bucket_delete_helper_exists():
 
 
 def test_app_version_v63():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v64_delete_clean_events_timeline.py

@@ -50,4 +50,4 @@ def test_timeline_has_stopped_state_without_running_animation_confusion():
 
 
 def test_app_version_v64():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v65_floating_form_timeline.py

@@ -57,4 +57,4 @@ def test_frontend_timeline_supports_stopped_status():
 
 
 def test_app_version_v65():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v66_clean_qwen_coder_path.py

@@ -19,4 +19,4 @@ def test_preferred_gpu_stays_zerogpu_only_and_fallback_a10g_large_default():
 
 
 def test_app_version_v66():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v67_monotonic_progress_events.py

@@ -36,4 +36,4 @@ def test_light_poll_cannot_reset_timeline_signature_to_earlier_steps():
 
 
 def test_app_version_v67():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v68_stable_events_timeline_centering.py

@@ -31,4 +31,4 @@ def test_timeline_centers_even_when_signature_is_unchanged():
 
 
 def test_app_version_v68():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v69_unified_progress_state.py

@@ -7,7 +7,7 @@ def test_current_step_uses_stabilized_timeline_not_view_header():
     assert "const stable=stabilizeProgressPayload(p,view)" in render
     assert "setText('currentStep',stable.currentStepLabel" in render
     assert "header.current_phase_label" not in render
-    assert "animateProgressBar(progress,runStatus,stable.currentStep)" in render
+    assert "animateProgressBar(progress,visualRunStatus,stable.currentStep)" in render
 
 
 def test_events_and_last_event_use_same_stable_event_source():
@@ -33,4 +33,4 @@ def test_timeline_centering_css_escape_is_safe():
 
 
 def test_app_version_v69():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v70_progress_resume_events_latency.py

@@ -58,4 +58,4 @@ def test_resume_latest_falls_back_to_latest_run():
 
 
 def test_app_version_v70():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v71_compact_stable_progress.py

@@ -32,4 +32,4 @@ def test_refresh_progress_prefers_canonical_status():
 
 
 def test_app_version_v71():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v72_clear_modes_space_link_gating.py

@@ -32,4 +32,4 @@ def test_launch_result_removed_and_selected_links_are_still_gated():
     assert "setLink('selectedOpenSettings',ready&&summary.target_space_url" in runs_js
 
 def test_app_version_v72():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v73_toast_timer_clean_active_run.py

@@ -27,4 +27,4 @@ def test_progress_poll_errors_use_floating_toast_not_internal_banner():
 
 
 def test_app_version_v73():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v74_grouped_timeline_rings.py

@@ -41,4 +41,4 @@ def test_timeline_dots_use_conic_gradient_ring():
 
 
 def test_app_version_v74():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v75_runs_sort_timeline_cost_note.py

@@ -38,4 +38,4 @@ def test_inference_provider_cost_note_is_visible_near_pi_model():
 
 
 def test_app_version_v75():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v76_coherent_progress_status_colors.py

@@ -50,4 +50,4 @@ def test_failed_events_superseded_by_later_step_are_shown_as_recovered_warning()
 
 
 def test_app_version_v76():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v77_terminal_report_refresh_no_launch_result.py

@@ -23,4 +23,4 @@ def test_terminal_run_still_refreshes_selected_run_detail_without_active_report_
 
 
 def test_app_version_v77():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v78_report_preview_cost_note_cleanup.py

@@ -36,4 +36,4 @@ def test_active_run_report_preview_block_removed():
 
 
 def test_app_version_v78():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v79_remove_active_report_preview.py

@@ -27,4 +27,4 @@ def test_selected_run_report_preview_support_can_remain_if_used_elsewhere():
 
 
 def test_app_version_v79():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v80_topbar_clean_timeline_failure_feedback.py

@@ -31,4 +31,4 @@ def test_recovered_failure_has_warning_color_not_green():
 
 
 def test_app_version_v80():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v81_robust_timeline_color_feedback.py

@@ -16,7 +16,7 @@ def test_timeline_prefers_problem_group_for_scroll_on_selected_old_runs():
     assert "function firstProblemGroupKey" in js
     assert "function currentCompactGroupKey(fullItems=[],overallStatus='')" in js
     assert "const problem=firstProblemGroupKey(fullItems,overallStatus)" in js
-    assert "const currentKey=currentCompactGroupKey(fullList,overallStatus)" in js
+    assert "const currentKey=trace.key||currentCompactGroupKey(fullList,overallStatus)" in js
 
 
 def test_timeline_scroll_resets_cleanly_between_selected_runs():
@@ -34,4 +34,4 @@ def test_recovered_failures_remain_amber_not_green():
 
 
 def test_app_version_v81():
-    assert '"version": "v81-robust-timeline-color-feedback"' in Path("app.py").read_text(encoding="utf-8")
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v82_elegant_failure_trace_timeline.py

@@ -0,0 +1,40 @@
+from pathlib import Path
+
+
+def test_timeline_has_explicit_failure_trace_note():
+    html = Path("web/index.html").read_text(encoding="utf-8")
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    css = Path("web/static/app.css").read_text(encoding="utf-8")
+    assert 'id="timelineTraceNote"' in html
+    assert "function timelineTraceModel" in js
+    assert "function renderTimelineTraceNote" in js
+    assert "Failure trace" in js
+    assert ".timeline-trace-note.failed" in css
+
+
+def test_failure_trace_styles_breakpoint_and_post_failure_tail():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    css = Path("web/static/app.css").read_text(encoding="utf-8")
+    assert "function traceClassForStep" in js
+    assert "trace-breakpoint" in js
+    assert "trace-post-failure" in js
+    assert "timeline.trace-failed .step.trace-breakpoint" in css
+    assert "repeating-linear-gradient" in css
+
+
+def test_old_runs_merge_events_before_stabilizing_timeline():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    render = js.split("function renderRunViewModel", 1)[1].split("function renderProgress", 1)[0]
+    assert render.index("mergeStableEvents") < render.index("stabilizeProgressPayload")
+    render_progress = js.split("function renderProgress", 1)[1]
+    assert render_progress.index("mergeStableEvents") < render_progress.index("stabilizeProgressPayload")
+
+
+def test_backend_timeline_failed_item_fallback_exists():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    assert "function failedItemInfoForGroup" in js
+    assert "failedItemInfoForGroup(group,byStep)" in js
+
+
+def test_app_version_v82():
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v83_success_progress_bar_color.py

@@ -0,0 +1,44 @@
+from pathlib import Path
+
+from src.progress import progress_from_events
+
+
+def test_frontend_success_detection_covers_real_success_statuses():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    components = Path("web/static/components.js").read_text(encoding="utf-8")
+    for token in [
+        "full_inference_candidate_health_passed",
+        "full_inference_success",
+        "health_only",
+        "repair_success",
+    ]:
+        assert token in js
+        assert token in components
+    assert "let klass=visualStatusClass(status);if(value>=100&&klass==='neutral')klass='success'" in js
+
+
+def test_progress_bar_prefers_backend_visual_status():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    assert "const visualRunStatus=p.visual_status||runStatus" in js
+    assert "animateProgressBar(progress,visualRunStatus" in js
+    assert "animateProgressBar(progress,p.visual_status||p.status" in js
+
+
+def test_backend_visual_status_success_for_health_only():
+    result = progress_from_events([
+        {"ts": "2026-06-02T10:00:00+00:00", "step": "done", "status": "health_only"}
+    ], state={"status": "health_only"})
+    assert result["status"] == "health_only"
+    assert result["visual_status"] == "success"
+    assert result["progress"] == 100
+
+
+def test_backend_visual_status_stopped_for_manual_or_blocker():
+    result = progress_from_events([
+        {"ts": "2026-06-02T10:00:00+00:00", "step": "done", "status": "manual_hardware_required"}
+    ], state={"status": "manual_hardware_required"})
+    assert result["visual_status"] == "stopped"
+
+
+def test_app_version_v83():
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v84_hardware_fallback_visual_state.py

@@ -0,0 +1,36 @@
+from pathlib import Path
+
+
+def test_hardware_fallback_has_dedicated_timeline_status_not_recovered_incident():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    assert "function isHardwareFallbackEvent" in js
+    assert "function isExpectedFallbackFailure" in js
+    assert "return 'fallback'" in js
+    assert "status==='fallback'" in js
+    assert "Hardware fallback" in js
+    assert "Recovered after an incident at Hardware" not in js
+
+
+def test_fallback_trace_message_is_neutral_routing_language():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    assert "ZeroGPU was not used for this run" in js
+    assert "switched to the fallback GPU" in js
+    assert "const title=trace.mode==='failed'?'Failure trace':(trace.mode==='fallback'?'Hardware fallback':'Recovered incident')" in js
+
+
+def test_hardware_fallback_has_blue_visual_style():
+    css = Path("web/static/app.css").read_text(encoding="utf-8")
+    assert ".step.status-fallback" in css
+    assert ".timeline-trace-note.fallback" in css
+    assert ".timeline.trace-fallback .step.trace-fallback" in css
+    assert ".activity-row.hardware-fallback" in css
+
+
+def test_activity_feed_labels_hardware_fallback_as_route():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    assert "return'hardware-fallback'" in js
+    assert "fallback route" in js
+
+
+def test_app_version_v84():
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v85_failure_reason_feedback.py

@@ -0,0 +1,41 @@
+from pathlib import Path
+
+
+def test_failure_and_fallback_reason_helpers_exist():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    assert "function eventReason" in js
+    assert "function zeroGpuReason" in js
+    assert "function failureReasonForGroup" in js
+    assert "function fallbackReasonForGroup" in js
+    assert "quota" in js
+    assert "ZeroGPU was unavailable" in js
+
+
+def test_timeline_trace_note_displays_reason():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    css = Path("web/static/app.css").read_text(encoding="utf-8")
+    assert "trace.reason" in js
+    assert '<b>Reason:</b>' in js
+    assert "trace-reason" in js
+    assert ".timeline-trace-note .trace-reason" in css
+
+
+def test_timeline_step_has_reason_tooltip_and_short_reason_line():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    css = Path("web/static/app.css").read_text(encoding="utf-8")
+    assert "reasonForTimelineItem" in js
+    assert "step-reason" in js
+    assert 'title="${escapeHtml(reason)}"' in js
+    assert ".step .step-reason" in css
+
+
+def test_activity_feed_can_show_event_reason_detail():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    css = Path("web/static/app.css").read_text(encoding="utf-8")
+    assert "event-reason" in js
+    assert "const reason=eventReason(item)" in js
+    assert ".activity-row .event-reason" in css
+
+
+def test_app_version_v85():
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v86_hf_error_detail_reasons.py

@@ -0,0 +1,42 @@
+from pathlib import Path
+
+
+def test_frontend_reads_error_reason_from_event_data_before_generic_message():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    assert "data.hf_error_detail" in js
+    assert "data.hf_error_message" in js
+    assert "data.response_text" in js
+    assert "data.response_body" in js
+    assert "data.error" in js
+    assert "data.reason" in js
+    assert "event.message" in js
+
+
+def test_zerogpu_reason_detects_quota_usage_exceeded_terms():
+    js = Path("web/static/progress.js").read_text(encoding="utf-8")
+    assert "lower.includes('quota')" in js
+    assert "lower.includes('usage')" in js
+    assert "lower.includes('exceeded')" in js
+    assert "ZeroGPU was unavailable" in js
+
+
+def test_worker_extracts_hf_http_exception_response_details():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert "def exception_payload" in worker
+    assert "response.text" in worker
+    assert "response.json()" in worker
+    assert '"hf_error_detail"' in worker
+    assert '"hf_error_message"' in worker
+    assert '"status_code"' in worker
+
+
+def test_space_hardware_events_include_hf_error_payload():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert "exc_payload = exception_payload(exc)" in worker
+    assert "create_repo_space_hardware" in worker
+    assert "**exc_payload" in worker
+    assert "Could not create Space with requested hardware" in worker
+
+
+def test_app_version_v86():
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v87_validation_panel_prefill_separation.py

@@ -0,0 +1,50 @@
+from pathlib import Path
+
+
+def test_validation_launch_does_not_replace_active_run():
+    js = Path("web/static/app.js").read_text(encoding="utf-8")
+    launch = js.split("async function launchValidate", 1)[1].split("async function refreshRunReportPreview", 1)[0]
+    assert "setActiveRun(result)" not in launch
+    assert "renderValidationLaunchResult(result)" in launch
+    assert "startValidationPolling(result.run_id)" in launch
+    assert "state.validationRunId" in js
+    assert "validationPoll" in js
+
+
+def test_validation_prefill_copies_args_kwargs_from_selected_run():
+    js = Path("web/static/app.js").read_text(encoding="utf-8")
+    assert "function validationPrefillFromDetail" in js
+    assert "smoke.test_args" in js
+    assert "launch.test_args_json" in js
+    assert "smoke.test_kwargs" in js
+    assert "launch.test_kwargs_json" in js
+    assert "setInputValue('testArgs',prettyJson(prefill.args))" in js
+    assert "setInputValue('testKwargs',prettyJson(prefill.kwargs))" in js
+
+
+def test_validation_form_enforces_json_array_and_object():
+    js = Path("web/static/app.js").read_text(encoding="utf-8")
+    app = Path("app.py").read_text(encoding="utf-8")
+    assert "Test args must be a JSON array" in js
+    assert "Test kwargs must be a JSON object" in js
+    assert "test_args_json must be a JSON array" in app
+    assert "test_kwargs_json must be a JSON object" in app
+
+
+def test_validation_ui_explains_separate_validation_run():
+    html = Path("web/index.html").read_text(encoding="utf-8")
+    css = Path("web/static/app.css").read_text(encoding="utf-8")
+    assert "Validation launches a separate validation run" in html
+    assert "does not replace the current Active Run" in html
+    assert ".validation-help" in css
+
+
+def test_default_validation_args_are_positional_not_inputs_object():
+    html = Path("web/index.html").read_text(encoding="utf-8")
+    test_args = html.split('id="testArgs"', 1)[1].split("</textarea>", 1)[0]
+    assert '"inputs"' not in test_args
+    assert "Explain the concept of agentic AI" in test_args
+
+
+def test_app_version_v87():
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v88_validation_schema_kwargs_sanitizer.py

@@ -0,0 +1,35 @@
+from pathlib import Path
+
+
+def test_validation_worker_extracts_endpoint_parameters_from_schema():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert "def endpoint_schema_for_api" in worker
+    assert "def endpoint_parameter_names" in worker
+    assert "parameter_name" in worker
+    assert "endpoint_parameters" in worker
+
+
+def test_validation_worker_sanitizes_unsupported_kwargs():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert "def sanitize_kwargs_for_schema" in worker
+    assert "safe_kwargs, dropped_kwargs, endpoint_parameters" in worker
+    assert "Ignored unsupported validation kwargs" in worker
+    assert "ignored_test_kwargs" in worker
+    assert "original_test_kwargs" in worker
+
+
+def test_validation_worker_retries_when_gradio_rejects_kwargs():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert "not a valid key-word argument" in worker
+    assert "Retrying validation without keyword arguments" in worker
+    assert "client.predict(*test_args, api_name=api_name)" in worker
+
+
+def test_validation_ui_explains_endpoint_specific_kwargs():
+    html = Path("web/index.html").read_text(encoding="utf-8")
+    assert "Test kwargs are endpoint-specific" in html
+    assert "ignores unsupported kwargs with a warning" in html
+
+
+def test_app_version_v88():
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v89_main_validation_zerogpu_contract_layout.py

@@ -0,0 +1,53 @@
+from pathlib import Path
+
+
+def test_space_test_panel_is_inside_workspace_after_active_run():
+    html = Path("web/index.html").read_text(encoding="utf-8")
+    active_idx = html.index('id="activeRunPanel"')
+    space_test_idx = html.index('id="spaceTestSection"')
+    runs_idx = html.index('id="runsExplorerPanel"')
+    workspace_end = html.index('</section>', runs_idx)
+    assert active_idx < space_test_idx < runs_idx
+    assert space_test_idx < workspace_end
+    css = Path("web/static/app.css").read_text(encoding="utf-8")
+    assert "v89: Space Test lives under Active Run" in css
+    assert "workspace-grid > .space-test-panel" in css
+    assert "grid-column:2" in css
+
+
+def test_build_job_passes_expected_output_type_to_worker():
+    app = Path("app.py").read_text(encoding="utf-8")
+    jobs = Path("src/jobs.py").read_text(encoding="utf-8")
+    assert "expected_output_type=payload.get(\"expected_output_type\")" in app
+    assert "expected_output_type: str | None = None" in jobs
+    assert 'env["EXPECTED_OUTPUT_TYPE"]' in jobs
+
+
+def test_main_worker_runs_live_generation_smoke_and_zerogpu_contract():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert "def run_generation_smoke" in worker
+    assert "Calling live generation endpoint for ZeroGPU timing" in worker
+    assert "measured_zero_gpu_recommendation" in worker
+    assert "recommendation_source" in worker
+    assert "recommendation_confidence" in worker
+    assert "observed_latency_seconds" in worker
+
+
+def test_inference_gate_uses_generation_smoke_success():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert "generation_smoke: dict | None" in worker
+    assert "generation_smoke_passed" in worker
+    assert "zero_gpu_duration_measured" in worker
+    assert "full_inference_success" in worker
+    assert "live generation smoke test passed" in worker.lower()
+
+
+def test_report_and_state_include_generation_smoke():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert '"generation_smoke": generation_smoke' in worker
+    assert "Live generation smoke / ZeroGPU duration" in worker
+    assert "zero_gpu_duration_recommendation" in worker
+
+
+def test_app_version_v89():
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v90_pi_model_mismatch_warning.py

@@ -0,0 +1,44 @@
+from pathlib import Path
+
+
+def test_worker_detects_pi_model_resolution_and_mismatch():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert "def detect_pi_model_resolution" in worker
+    assert "def extract_pi_models_from_text" in worker
+    assert "Kimi" in worker
+    assert "pi_model_resolution.json" in worker
+    assert "mismatch" in worker
+    assert "emit_pi_model_resolution" in worker
+
+
+def test_worker_emits_pi_model_mismatch_event_and_state():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert "Pi assistant model appears to differ from the requested model" in worker
+    assert '"pi_model_resolution": pi_model_resolution' in worker
+    assert "## Pi assistant model" in worker
+
+
+def test_bucket_exposes_pi_model_resolution_to_ui():
+    bucket = Path("src/bucket.py").read_text(encoding="utf-8")
+    assert "pi_model_resolution.json" in bucket
+    assert '"pi_model_resolution"' in bucket
+
+
+def test_ui_warns_when_pi_provider_model_changed():
+    runs = Path("web/static/runs.js").read_text(encoding="utf-8")
+    progress = Path("web/static/progress.js").read_text(encoding="utf-8")
+    css = Path("web/static/app.css").read_text(encoding="utf-8")
+    assert "Pi assistant model changed" in runs
+    assert "Inference Providers routing/fallback" in runs
+    assert "pi_model_resolution" in progress
+    assert "v90: Pi/provider model mismatch warning" in css
+
+
+def test_pi_model_resolution_step_is_in_timeline():
+    progress = Path("web/static/progress.js").read_text(encoding="utf-8")
+    assert "pi_model_resolution" in progress
+    assert "Pi verify" in progress
+
+
+def test_app_version_v90():
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")

tests/test_v91_pi_model_detection_from_published_traces.py

@@ -0,0 +1,29 @@
+from pathlib import Path
+
+
+def test_bucket_can_reconstruct_pi_model_resolution_from_published_traces():
+    bucket = Path("src/bucket.py").read_text(encoding="utf-8")
+    assert "def _trace_based_pi_model_resolution" in bucket
+    assert "traces/redacted/**/*.jsonl" in bucket
+    assert "logs/pi_output.txt" in bucket
+    assert "published_pi_traces" in bucket
+    assert "_extract_pi_models_from_text" in bucket
+
+
+def test_read_run_bundle_uses_trace_fallback_when_resolution_file_missing():
+    bucket = Path("src/bucket.py").read_text(encoding="utf-8")
+    assert 'if not bundle.get("pi_model_resolution")' in bucket
+    assert '_trace_based_pi_model_resolution(run_id, bundle' in bucket
+    assert '"pi_model_resolution": pi_model_resolution' in bucket
+
+
+def test_worker_marks_model_resolution_source_as_after_trace_sync():
+    worker = Path("src/worker_payload.py").read_text(encoding="utf-8")
+    assert "Pi sessions are the source of truth" in worker
+    assert "published_pi_traces_after_sync" in worker
+    assert "collect_pi_traces(run_dir, events_path)" in worker
+    assert "detect_pi_model_resolution(pi_model, run_dir, pi_out)" in worker
+
+
+def test_app_version_v91():
+    assert '"version": "v91-pi-model-detection-from-published-traces"' in Path("app.py").read_text(encoding="utf-8")