argus-testing 0.12.2__tar.gz → 0.13.0__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (212) hide show
  1. {argus_testing-0.12.2 → argus_testing-0.13.0}/.gitignore +1 -0
  2. {argus_testing-0.12.2 → argus_testing-0.13.0}/AGENTS.md +13 -4
  3. {argus_testing-0.12.2 → argus_testing-0.13.0}/CHANGELOG.md +179 -0
  4. {argus_testing-0.12.2 → argus_testing-0.13.0}/CONTRIBUTING.md +22 -0
  5. {argus_testing-0.12.2 → argus_testing-0.13.0}/PKG-INFO +9 -6
  6. {argus_testing-0.12.2 → argus_testing-0.13.0}/README.md +8 -5
  7. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/_compat.py +3 -1
  8. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/access.py +4 -1
  9. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/journal.py +2 -0
  10. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/monitoring.py +4 -5
  11. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/reporting/common.py +14 -2
  12. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/reporting/html.py +10 -2
  13. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/reporting/junit.py +16 -6
  14. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/reporting/sarif.py +10 -2
  15. argus_testing-0.13.0/argus/adapters/scripts/collect_text_js.js +136 -0
  16. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/scripts/extract_elements_js.js +53 -5
  17. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/scripts/extract_page_content_js.js +21 -12
  18. argus_testing-0.13.0/argus/adapters/scripts/ref_engine_js.js +16 -0
  19. argus_testing-0.13.0/argus/adapters/scripts/row_text_js.js +26 -0
  20. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/actions.py +43 -28
  21. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/coverage.py +16 -7
  22. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/evidence.py +7 -3
  23. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/expectations.py +7 -5
  24. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/exports.py +10 -0
  25. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/findings.py +4 -4
  26. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/inspection.py +4 -1
  27. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/journeys.py +40 -5
  28. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/lifecycle.py +20 -12
  29. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/manager.py +9 -1
  30. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/navigation.py +4 -1
  31. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/network_storage.py +15 -0
  32. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/observation.py +13 -6
  33. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/probes.py +88 -27
  34. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/regression.py +54 -6
  35. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/screen.py +110 -59
  36. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/screenshots.py +8 -7
  37. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/testgen.py +178 -81
  38. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/text.py +54 -52
  39. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/verification.py +89 -57
  40. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/browser.py +218 -64
  41. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/cleanup.py +10 -2
  42. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/cli.py +29 -1
  43. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/differ.py +5 -3
  44. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/domain/models.py +22 -0
  45. argus_testing-0.13.0/argus/domain/verdict.py +345 -0
  46. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/domain/verification.py +2 -12
  47. argus_testing-0.13.0/argus/domain/wordchars.py +6 -0
  48. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/mcp_server.py +9 -2
  49. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/ports.py +2 -1
  50. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/resolver.py +64 -22
  51. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/screen/backend.py +250 -106
  52. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/screen/safety.py +29 -17
  53. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/security.py +27 -15
  54. {argus_testing-0.12.2 → argus_testing-0.13.0}/docs/agent-protocol.md +78 -16
  55. {argus_testing-0.12.2 → argus_testing-0.13.0}/docs/architecture.md +7 -4
  56. {argus_testing-0.12.2 → argus_testing-0.13.0}/docs/decisions.md +19 -0
  57. {argus_testing-0.12.2 → argus_testing-0.13.0}/docs/evaluation.md +1 -1
  58. {argus_testing-0.12.2 → argus_testing-0.13.0}/docs/usage.md +24 -6
  59. {argus_testing-0.12.2 → argus_testing-0.13.0}/examples/response-overhead.json +3 -3
  60. {argus_testing-0.12.2 → argus_testing-0.13.0}/llms-install.md +7 -2
  61. {argus_testing-0.12.2 → argus_testing-0.13.0}/pyproject.toml +1 -1
  62. {argus_testing-0.12.2 → argus_testing-0.13.0}/server.json +2 -2
  63. argus_testing-0.13.0/tests/test_action_time_resolution.py +159 -0
  64. argus_testing-0.13.0/tests/test_audit_hardening.py +276 -0
  65. argus_testing-0.13.0/tests/test_crawl_site.py +85 -0
  66. argus_testing-0.13.0/tests/test_element_identity.py +154 -0
  67. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_evidence_provenance.py +21 -0
  68. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_export_guards.py +34 -4
  69. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_export_privacy.py +20 -0
  70. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_export_test.py +180 -1
  71. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_final_hardening.py +4 -0
  72. argus_testing-0.13.0/tests/test_frame_extraction.py +116 -0
  73. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_mcp_contract.py +24 -0
  74. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_monitoring.py +10 -1
  75. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_release_metadata.py +3 -0
  76. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_reproduction_receipt.py +4 -2
  77. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_resolver.py +107 -0
  78. argus_testing-0.13.0/tests/test_row_textarea.py +127 -0
  79. argus_testing-0.13.0/tests/test_screen_errors.py +154 -0
  80. argus_testing-0.13.0/tests/test_screen_input_guard.py +194 -0
  81. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_security_boundaries.py +21 -0
  82. argus_testing-0.13.0/tests/test_textarea_label.py +85 -0
  83. argus_testing-0.13.0/tests/test_verdict.py +177 -0
  84. argus_testing-0.13.0/tests/test_verdict_integrity.py +365 -0
  85. argus_testing-0.13.0/tests/test_verdict_parity.py +405 -0
  86. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_verification_contract.py +2 -2
  87. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_verification_efficiency.py +33 -0
  88. {argus_testing-0.12.2 → argus_testing-0.13.0}/LICENSE +0 -0
  89. {argus_testing-0.12.2 → argus_testing-0.13.0}/PRIVACY.md +0 -0
  90. {argus_testing-0.12.2 → argus_testing-0.13.0}/SECURITY.md +0 -0
  91. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/__init__.py +0 -0
  92. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/__init__.py +0 -0
  93. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/assets/__init__.py +0 -0
  94. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/assets/dashboard.css +0 -0
  95. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/assets/dashboard.html +0 -0
  96. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/assets/dashboard.js +0 -0
  97. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/logging.py +0 -0
  98. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/reporting/__init__.py +0 -0
  99. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/reporting/json.py +0 -0
  100. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/reporting/privacy.py +0 -0
  101. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/reporting/reporter.py +0 -0
  102. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/scripts/__init__.py +0 -0
  103. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/scripts/check_layout_js.js +0 -0
  104. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/scripts/extract_performance_js.js +0 -0
  105. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/scripts/find_visible_element_js.js +0 -0
  106. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/adapters/scripts/inspect_element_js.js +0 -0
  107. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/__init__.py +0 -0
  108. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/capsules.py +0 -0
  109. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/capture.py +0 -0
  110. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/context.py +0 -0
  111. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/events.py +0 -0
  112. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/finish.py +0 -0
  113. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/guidance.py +0 -0
  114. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/operations.py +0 -0
  115. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/payloads.py +0 -0
  116. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/progress.py +0 -0
  117. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/session.py +0 -0
  118. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/telemetry.py +0 -0
  119. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/application/verifiers.py +0 -0
  120. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/bootstrap.py +0 -0
  121. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/detector.py +0 -0
  122. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/doctor.py +0 -0
  123. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/domain/__init__.py +0 -0
  124. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/domain/contracts.py +0 -0
  125. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/domain/coverage.py +0 -0
  126. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/domain/detector.py +0 -0
  127. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/domain/findings.py +0 -0
  128. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/domain/replay.py +0 -0
  129. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/entrypoints/__init__.py +0 -0
  130. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/entrypoints/mcp/__init__.py +0 -0
  131. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/entrypoints/mcp/catalog.py +0 -0
  132. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/entrypoints/mcp/compat.py +0 -0
  133. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/entrypoints/mcp/overhead.py +0 -0
  134. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/entrypoints/mcp/results.py +0 -0
  135. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/entrypoints/mcp/server.py +0 -0
  136. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/entrypoints/mcp/snapshots.py +0 -0
  137. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/install_check.py +0 -0
  138. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/measurement.py +0 -0
  139. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/models.py +0 -0
  140. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/reporter.py +0 -0
  141. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/runtime.py +0 -0
  142. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/screen/__init__.py +0 -0
  143. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/screen/permissions.py +0 -0
  144. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/screen/validate.py +0 -0
  145. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/settings.py +0 -0
  146. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus/storage.py +0 -0
  147. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/__init__.py +0 -0
  148. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/bench/__init__.py +0 -0
  149. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/bench/__main__.py +0 -0
  150. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/bench/runner.py +0 -0
  151. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/bench/scenarios_buggytasks.py +0 -0
  152. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/bench/scenarios_darkshop.py +0 -0
  153. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/__init__.py +0 -0
  154. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/__main__.py +0 -0
  155. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/adjudication.py +0 -0
  156. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/agent_fixture.py +0 -0
  157. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/browser_baseline.py +0 -0
  158. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/codex_runner.py +0 -0
  159. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/fixture.py +0 -0
  160. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/gateway.py +0 -0
  161. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/models.py +0 -0
  162. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/overhead.py +0 -0
  163. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/scoring.py +0 -0
  164. {argus_testing-0.12.2 → argus_testing-0.13.0}/argus_lab/evaluation/smoke.py +0 -0
  165. {argus_testing-0.12.2 → argus_testing-0.13.0}/docs/evaluation-0.8.md +0 -0
  166. {argus_testing-0.12.2 → argus_testing-0.13.0}/examples/agent-evaluation-optimized-evidence.json +0 -0
  167. {argus_testing-0.12.2 → argus_testing-0.13.0}/examples/agent-evaluation-optimized.json +0 -0
  168. {argus_testing-0.12.2 → argus_testing-0.13.0}/examples/agent-evaluation-pilot-evidence.json +0 -0
  169. {argus_testing-0.12.2 → argus_testing-0.13.0}/examples/agent-evaluation-pilot.json +0 -0
  170. {argus_testing-0.12.2 → argus_testing-0.13.0}/examples/mcp-config.json +0 -0
  171. {argus_testing-0.12.2 → argus_testing-0.13.0}/glama.json +0 -0
  172. {argus_testing-0.12.2 → argus_testing-0.13.0}/human-eye-fixture/app.py +0 -0
  173. {argus_testing-0.12.2 → argus_testing-0.13.0}/scripts/check_distributions.py +0 -0
  174. {argus_testing-0.12.2 → argus_testing-0.13.0}/test-site/app.py +0 -0
  175. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/__init__.py +0 -0
  176. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/conftest.py +0 -0
  177. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_action_targeting.py +0 -0
  178. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_adjudication.py +0 -0
  179. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_agent_results.py +0 -0
  180. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_agent_workflow.py +0 -0
  181. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_architecture_runtime.py +0 -0
  182. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_bench_report.py +0 -0
  183. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_browser_shadow.py +0 -0
  184. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_codex_evaluation.py +0 -0
  185. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_cross_stack.py +0 -0
  186. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_deprecated_modules.py +0 -0
  187. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_detector_console.py +0 -0
  188. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_detector_network.py +0 -0
  189. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_distribution_contents.py +0 -0
  190. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_doctor.py +0 -0
  191. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_evaluation.py +0 -0
  192. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_expectation_oracle.py +0 -0
  193. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_final_coverage.py +0 -0
  194. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_form_checkboxes.py +0 -0
  195. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_journal_regression.py +0 -0
  196. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_mcp_compat.py +0 -0
  197. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_real_app_interactions.py +0 -0
  198. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_regression_cli.py +0 -0
  199. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_replay_minimize.py +0 -0
  200. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_replay_receipt.py +0 -0
  201. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_replay_trace.py +0 -0
  202. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_reporter_badge.py +0 -0
  203. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_response_efficiency.py +0 -0
  204. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_runtime_limits.py +0 -0
  205. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_screen_permissions.py +0 -0
  206. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_screen_safety.py +0 -0
  207. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_session_lifecycle.py +0 -0
  208. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_state_capsules.py +0 -0
  209. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_stdio_protocol.py +0 -0
  210. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_storage_cleanup.py +0 -0
  211. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_substrate.py +0 -0
  212. {argus_testing-0.12.2 → argus_testing-0.13.0}/tests/test_visual_inspection.py +0 -0
@@ -24,3 +24,4 @@ docs/*
24
24
  !docs/evaluation.md
25
25
  !docs/evaluation-0.8.md
26
26
  !docs/usage.md
27
+ .claude/
@@ -84,7 +84,7 @@ result includes captured stages. See [the protocol](docs/agent-protocol.md).
84
84
  | `test_action(target, expect=...)` | Click + before/after diff in one call. Pass `expect` to PREDICT the outcome ({"count":{"label":"tasks","delta":1}}, {"gains":"Buy milk"}, {"removes":...}, {"text_present":...}, {"toast":...}, {"url_changed":true}) and Argus reports MATCH / SURPRISE — a surprise is a bug lead. Also shows CROSS-STACK: which requests the click fired (methods/statuses) and a CHECK nudge when a message appeared without a matching write. |
85
85
  | `verify_persistence(expect, target_text, after_url)` | Forces a fresh GET; returns target presence and the page observation. Check persistence once, repeat only to resolve uncertainty. A matching symptom is not proof of a semantic defect. |
86
86
  | `capsule_save(name, liveness_marker)` / `capsule_restore(name)` | Snapshot cookies and page storage after establishing state through the UI. Restore replaces the captured client state and checks a saved liveness marker when provided; without one, validity remains unverified. It does not reset backend data or guarantee identical application state across journeys. |
87
- | `regression_check()` | Re-test the findings journaled in prior runs against the CURRENT build: STILL-PRESENT / NO-LONGER-REPRODUCES / INCONCLUSIVE. "Did my fix land, did anything come back?" |
87
+ | `regression_check()` | Re-test journaled findings against the current build: `STILL-PRESENT`, `FIXED` (the recorded symptom no longer reproduces), or `INCONCLUSIVE`. Mixed receipts remain inconclusive; a `FIXED` result is not proof of application correctness. |
88
88
  | `export_test(finding_id, run=False, save_private_values=False)` | Export a standalone Playwright test from a finding verified by a fresh-load text check. Confirm it fails on the symptom before using it in CI. Unsupported steps are marked TODO; 0.12.1 blocks incomplete exports before execution. See the [export protocol](docs/agent-protocol.md#exporting-a-regression-test) for environment variables and package-version limits. `run=true` repeats the journey's writes, so use it only with the user's authority. Only set `save_private_values=true` when the user explicitly requests retaining local replay values; its returned JSON file can contain credentials and must never be published. |
89
89
  | `inspect_element(description)` | Computed styles + ARIA + outerHTML + truncation flag for interactive or visible non-interactive content. |
90
90
  | `check_layout()` | Bounded viewport signals for horizontal overflow, clipped text, small targets, and fixed/sticky overlays. |
@@ -98,7 +98,7 @@ result includes captured stages. See [the protocol](docs/agent-protocol.md).
98
98
  | `crawl_site()` | Page discovery: crawls internal links, auto-capturing only console/network events (tagged). Walk the surfaced pages and record_bug what you confirm. |
99
99
  | `end_session(coverage_updates=...)` | Optionally submit explicit final exercised/blocked goals with evidence, then close and export reports. The entire batch validates before shutdown; invalid updates keep the session open. |
100
100
 
101
- ### Screen mode (macOS)
101
+ ### Screen mode (macOS, experimental)
102
102
 
103
103
  | Tool | Purpose |
104
104
  |------|---------|
@@ -221,8 +221,11 @@ Don't record speculation, polish nits, or static a11y / SEO that
221
221
  - Session cap: 30 min default
222
222
  (`ARGUS_SCREEN_SESSION_MAX_SECONDS`).
223
223
  - Panic button: `touch ~/.argus/abort` blocks every subsequent screen
224
- action in the current session
225
- (`ARGUS_SCREEN_ABORT_FILE` to relocate).
224
+ action, including `screen_launch` and `screen_quit`, and no new screen
225
+ session starts while the file exists. Only the user removes it; never
226
+ delete it or work around it (`ARGUS_SCREEN_ABORT_FILE` to relocate).
227
+ - Keyboard and coordinate input is sent only while the bound app is
228
+ frontmost. Argus brings it forward first and refuses if that fails.
226
229
  - Every `screen_click_what` / `screen_type_into` writes a
227
230
  before-and-after screenshot pair to the action trail.
228
231
 
@@ -251,6 +254,12 @@ application data.
251
254
  journal), 1 for still-present findings, 2 for operational/storage failures and
252
255
  3 for inconclusive checks. An empty journal is not proof of app correctness.
253
256
 
257
+ Since 0.12.2, automatic regression refuses storage-reset checks and legacy entries
258
+ without complete preconditions. Preserve journals from 0.12.1 and earlier and
259
+ re-record findings in a new output directory. Regenerate older exported tests;
260
+ see [migration guidance](docs/usage.md#regression-gates-in-ci). Never drop a
261
+ storage-reset precondition to make a finding exportable.
262
+
254
263
  ## What changed in 0.5 (if you knew 0.4)
255
264
 
256
265
  - `get_page_state` → `observe` (no integer indices).
@@ -1,5 +1,184 @@
1
1
  # Changelog
2
2
 
3
+ ## Plugin updates - 2026-11-10
4
+
5
+ - Claude plugin 0.12.4 and Codex plugin 0.12.3 pin the published Argus 0.12.3
6
+ runtime and refresh their dependency locks and setup commands.
7
+
8
+ ## 0.13.0 - 2026-11-12
9
+
10
+ One set of verdict rules. Live verification, `verify_persistence` and exported
11
+ tests decided "is the symptom present?" with three separate implementations
12
+ that had drifted apart; they now run the same code. No new tools.
13
+
14
+ Upgrading: regenerate exported tests to get the shared rules (earlier files
15
+ keep working with their old rules). "Present" now means text a user can see or
16
+ the value of a visible field, so text reachable only through `aria-label`,
17
+ unselected `<option>` text and closed `<details>` content no longer count.
18
+
19
+
20
+ - Description-driven web tools (`click_what`, `type_into`, `test_action`,
21
+ `inspect_element` and the rest) resolve the description against the page as
22
+ it is when the action runs, not against the last observation: a row that
23
+ re-rendered in place is found by its current text, and a description that has
24
+ become ambiguous or no longer matches returns candidates or a no-match error
25
+ instead of acting.
26
+ - Screen-mode tools report failures and refusals as errors with a code
27
+ (`permission_denied` for safety refusals and missing macOS grants, `timeout`,
28
+ `invalid_argument`, `backend_error`) instead of completed calls.
29
+ - A `<textarea>` is named by its aria-label, `<label>`, placeholder or name, and
30
+ its initial content is reported only as its value, so a secret-named field's
31
+ content is hidden as `(value hidden)` instead of shown as its label.
32
+ - Row text separates adjacent elements that the markup writes without
33
+ whitespace, as frameworks usually do. `<span>Buy groceries</span><button>Delete</button>`
34
+ read as "Buy groceriesDelete", so "Delete near Buy groceries" matched nothing.
35
+ - A row's text (the `near:` context, repeating item lists and exported tests'
36
+ row matching) leaves out `<textarea>` content, so a secret in a textarea no
37
+ longer shows through the row of every control beside it. Dialog previews,
38
+ toasts and the focused element follow the same rule, and the focused
39
+ element's value is no longer collected.
40
+ - Fresh-load verification and `verify_persistence` read the same visible text:
41
+ the page and every visible frame (cross-origin frames and open shadow roots
42
+ included) plus the values of visible fields. Text that is only inside an
43
+ iframe is now found, and hidden text inside a visible list row or element no
44
+ longer counts as present.
45
+ - Login-wall phrases are searched in the main frame's visible text, so a sign-in
46
+ banner rendered in a shadow root makes an absent symptom inconclusive.
47
+ - `verify_persistence` suspends the session's network mocks for its fresh load
48
+ and restores them afterwards, so it can no longer confirm a response the agent
49
+ injected, and reports an absent target on a page whose network never settles
50
+ as inconclusive.
51
+ - Exported regression tests embed Argus's verdict rules and visible-text
52
+ collector verbatim (the docstring names the rules version), so they decide a
53
+ symptom exactly as fresh-load verification and `verify_persistence` do:
54
+ unspaced scripts such as CJK match, hidden field values do not count, open
55
+ shadow roots are read, and a redirect elsewhere or a non-English login form
56
+ makes an absent symptom inconclusive.
57
+ - `export_test(run=True)` reports a new outcome, `inconclusive`, with a reason
58
+ code, for an error page, login wall, redirect, unsettled or unreadable page;
59
+ these were `errored` before. The file's last stderr line starts with
60
+ `ARGUS_INCONCLUSIVE`; the test fails under pytest and the file exits 3 when
61
+ run as a script. Earlier exports keep their old rules until regenerated.
62
+ - A child frame that cannot be read now gives its own inconclusive reason
63
+ instead of claiming the page text exceeded the inspection limit.
64
+ - Capsule liveness markers, `test_action`'s `text_present`/`text_absent` and the
65
+ action diff read the same visible frame text.
66
+
67
+ ## 0.12.3 - 2026-11-10
68
+
69
+ Fixes from a full audit of 0.12.2. Journals and exported tests stay compatible.
70
+ Some checks that returned `FIXED` or `verified` on evidence that could not show
71
+ the symptom now return inconclusive, and failed actions are reported as errors.
72
+ Screen mode is experimental.
73
+
74
+ - Check symptom text against the whole page instead of its first 5,000
75
+ characters, so text further down is no longer reported as gone. Match targets
76
+ inside Chinese, Japanese, Korean and Thai text. Keep lists that share a class
77
+ or tag apart.
78
+ - Load a new document for fresh-load checks of hash-route URLs; a same-document
79
+ navigation previously read in-memory state.
80
+ - Treat an HTTP-status check as inconclusive when the load answers with a
81
+ different error, is redirected or lands on a login page. Only a clean answer
82
+ from the requested page counts as no longer reproduced.
83
+ - Treat an absent symptom as inconclusive when the load ended on another page
84
+ or shows a login form in any common UI language.
85
+ - `argus-regression` exits 2 for a URL without a scheme or with credentials and
86
+ for journal entries of the wrong type, and names other origins that have
87
+ journals when the given one has none.
88
+ - Report intermittent receipts as inconclusive in JUnit (a failure, labelled
89
+ INTERMITTENT), SARIF and the HTML badge. Map SARIF `critical` to `error`.
90
+ Strip characters XML cannot carry. Show inconclusive checks as INCONCLUSIVE in
91
+ HTML. Redact regression artifacts.
92
+ - `argus-cleanup` ignores report files whose name is not a run ID; one such
93
+ name could select the whole output directory. Automatic screenshot names can
94
+ no longer leave the run directory.
95
+ - Make credential redaction linear in input size and cap captured request
96
+ bodies. Stop redacting ordinary keys such as `author` and `sidebar`; redact
97
+ `pwd`, quoted values, full Bearer and Basic values, signatures and private keys.
98
+ - Observe forms that contain a control named `id`. Attach page listeners once.
99
+ Accept unsaved-changes prompts so navigation proceeds. Recover after a failed
100
+ navigation. Judge internal links by origin.
101
+ - Act on the exact element that was listed. Each observed element keeps a
102
+ reference to its DOM node, so a click or fill no longer lands on an earlier
103
+ control that shares a label prefix, on a hidden hover-only control in another
104
+ row, or on a same-named field in another form. An element that has been
105
+ removed fails the action instead of redirecting it.
106
+ - `test_form` with automatic submit only considers controls in the form that
107
+ was filled.
108
+ - Screen mode: send keyboard and coordinate input only while the bound app is
109
+ frontmost; a session started without a target stays with its first app.
110
+ The abort file now also stops `screen_launch` and `screen_quit`, and a new
111
+ session no longer deletes it: only the user clears it. `screen_quit` needs an
112
+ exact or unique app match. Hold, drag and click counts are bounded, key
113
+ combinations apply modifiers per key and accept hyphenated key names, and
114
+ failed typing no longer echoes the text. `screen_wait_for_stable` works.
115
+ `ARGUS_SCREEN_ABORT_FILE` expands `~`, and its directory is created at start.
116
+ - `test_action` and `test_form` return the console and HTTP failures an action
117
+ raised, and the cross-stack check, as structured data (`browser_events`,
118
+ `cross_stack`); compact and delta responses previously dropped them.
119
+ - Failed actions (hover, resize, upload, drag, coordinate input, missing
120
+ screenshots, a screen session that cannot start) are reported as errors
121
+ instead of `status: completed`.
122
+ - `argus-regression` gives each finding its own deadline; a long journal or one
123
+ unreachable page no longer fails the whole run.
124
+ - Coverage goals match on whole words.
125
+ - Network mocks are registered on the browser context with one handler per
126
+ pattern. Replacing a mock no longer leaves the first response active during
127
+ a fresh-load verification, and mocks follow tab switches and device emulation.
128
+ - CI also runs on Python 3.13 and 3.14.
129
+ - Screen mode reads the frontmost app from the accessibility and window
130
+ servers. `NSWorkspace` is only refreshed by a running main run loop, so a
131
+ long-lived server kept seeing the app that was frontmost at startup and
132
+ refused all input. `screen_quit` and `screen_launch` report a name that
133
+ matches several running apps instead of "not running".
134
+ - A start rejected after launch closes the browser; `end_session` always leaves
135
+ the failed state. `test_form` no longer echoes a typed value when a fill fails.
136
+ - Resolve "Last name" and "First name" to the fields with those labels instead
137
+ of reading the word as a position. Only the last kind word sets the kind, so
138
+ "Select all button" and "Menu button" pick the button. A bare label shared by
139
+ several rows stays ambiguous. Chinese, Japanese, Korean and Thai descriptions
140
+ match inside longer labels.
141
+ - List elements in `srcdoc` and blank iframes once, and reach controls in
142
+ cross-origin frames nested inside other cross-origin frames.
143
+ - An exported step located by a structural CSS path (`:nth-of-type`,
144
+ `:nth-child`) is positional: the export is incomplete and does not run.
145
+ - `drag_what`, `right_click`, `storage_set`, `storage_remove`, `storage_clear`,
146
+ `cookies_set`, `cookies_clear`, `tabs_switch` and `eval_js` appear in the
147
+ action trace without their values, so an export that contains them is
148
+ incomplete and a replay through them reports that it cannot repeat them.
149
+ - `crawl_site` stays on the origin it started from, skips a link that redirects
150
+ elsewhere, returns to the start page, and counts only the event-bugs it captured.
151
+ - `network_mock`, `network_unmock`, `network_clear_mocks`, `tabs_close` and
152
+ `set_dialog_handler` appear in the action trace, the mock without its body and
153
+ the dialog handler without its text, so an export after a mock is incomplete.
154
+ - A recheck that verifies a finding recorded without a goal updates the coverage
155
+ evidence of every goal whose window holds it, not only of tagged goals.
156
+ - Element descriptions hide the value of a text field whose name, id, label or
157
+ placeholder names a secret, such as `api_key`.
158
+ - `wait_for_request` redacts credentials in the matched URL.
159
+ - The monitor rejects same-site requests, such as an image on the app under test
160
+ pointing at the event stream; only same-origin, direct and non-browser
161
+ requests are accepted.
162
+ - An empty `ARGUS_TOOL_PROFILE` means `web`, and an invalid one is reported as a
163
+ usage error naming the valid profiles instead of a traceback.
164
+
165
+ ## Documentation and Glama deployment - 2026-11-09
166
+
167
+ - Synchronize the agent installation guide and protocol with Argus 0.12.2,
168
+ including journal migration, regression verdicts and export preconditions.
169
+ - Publish Glama 0.12.2 from release commit
170
+ `e2ac33da3b3eda998c3ba98084c33ed291381b66` after adding `libatomic1` to its
171
+ build steps. The verified build reports MCP server version 0.12.2.
172
+ - Document the system dependency and the wait between build success and image
173
+ availability for release creation. The runtime package version is unchanged.
174
+
175
+ ## Plugin updates - 2026-11-08
176
+
177
+ - Claude plugin 0.12.3 and Codex plugin 0.12.2 pin the published Argus 0.12.2
178
+ runtime and refresh their dependency locks and setup commands.
179
+ - Regenerate older exported tests to use the regression reliability fixes;
180
+ keep legacy journals and re-record findings in a new report directory.
181
+
3
182
  ## 0.12.2 - 2026-11-08
4
183
 
5
184
  - Treat intermittent regression receipts as inconclusive instead of fixed.
@@ -90,6 +90,11 @@ For release, synchronize `pyproject.toml` and both version fields in
90
90
  `server.json`, update `CHANGELOG.md`, build with `python -m build`, check with
91
91
  `twine check`, and smoke-test the wheel outside the source directory. JS and
92
92
  dashboard assets are package resources and must be present in the wheel.
93
+ Update the current-release section in `README.md` and the version and install pin
94
+ in `llms-install.md`. When behavior changes, align `docs/usage.md`,
95
+ `docs/agent-protocol.md`, architecture and privacy guidance with the implemented
96
+ contract. Keep historical version notes and evaluation results tied to the
97
+ versions they actually describe.
93
98
  Source distributions use an explicit inclusion list. Run
94
99
  `python scripts/check_distributions.py dist/*` after building; newly packaged
95
100
  files must satisfy the public-content policy before publication.
@@ -106,6 +111,23 @@ publication succeed, manually publish that verified Glama build under the same
106
111
  version. A version label does not prove which source was built. Do not reuse an
107
112
  occupied version label for a different build.
108
113
 
114
+ For Glama's `debian:bookworm-slim` build, install `libatomic1` before invoking
115
+ Playwright. Its bundled Node driver needs `libatomic.so.1` to start, including
116
+ when running `playwright install --with-deps`. Use these build steps:
117
+
118
+ ```json
119
+ [
120
+ "apt-get update && apt-get install -y --no-install-recommends libatomic1",
121
+ "uv sync",
122
+ "uv run playwright install --with-deps chromium"
123
+ ]
124
+ ```
125
+
126
+ A successful build can precede the image becoming available in Glama's registry.
127
+ If release creation reports that the image is not available yet, wait and retry
128
+ the same verified build. Confirm the published release links to that build;
129
+ do not treat a successful test alone as a published release.
130
+
109
131
  After publication, verify the PyPI wheel and source archive, official MCP
110
132
  Registry package version, and directory release associations. Run
111
133
  `python -m argus.install_check` against the published package outside the checkout.
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.5
2
2
  Name: argus-testing
3
- Version: 0.12.2
3
+ Version: 0.13.0
4
4
  Summary: Testing MCP for coding agents: exercise real user journeys, find bugs, and verify them with evidence and persistence checks
5
5
  Project-URL: Homepage, https://yichenwu.dev/projects/argus
6
6
  Project-URL: Documentation, https://github.com/chriswu727/argus#readme
@@ -129,11 +129,14 @@ replaying a journey can repeat its writes.
129
129
  [Cursor, other MCP clients and troubleshooting](docs/usage.md#quick-start) ·
130
130
  [Claude plugin](plugins/claude) · [Codex plugin](plugins/codex)
131
131
 
132
- **0.12.2**: regression checks preserve complete verification conditions and treat
133
- intermittent results as inconclusive. Exported tests stop on missing or ambiguous
134
- row targets. Regenerate older tests and follow the [journal migration steps](docs/usage.md).
135
- See the [releases](https://github.com/chriswu727/argus/releases) and
136
- [changelog](CHANGELOG.md#0122---2026-11-08).
132
+ **0.13.0**: live verification, `verify_persistence` and exported tests now
133
+ decide with the same rules, checked against each other on every change. They
134
+ read what a user can see, in every frame and open shadow root, and an exported
135
+ test reports an inconclusive check instead of a false failure. Regenerate
136
+ exported tests to get the shared rules; see the
137
+ [journal migration steps](docs/usage.md#regression-gates-in-ci), the
138
+ [releases](https://github.com/chriswu727/argus/releases) and the
139
+ [changelog](CHANGELOG.md#0130---2026-11-12).
137
140
 
138
141
  ## Engineering
139
142
 
@@ -85,11 +85,14 @@ replaying a journey can repeat its writes.
85
85
  [Cursor, other MCP clients and troubleshooting](docs/usage.md#quick-start) ·
86
86
  [Claude plugin](plugins/claude) · [Codex plugin](plugins/codex)
87
87
 
88
- **0.12.2**: regression checks preserve complete verification conditions and treat
89
- intermittent results as inconclusive. Exported tests stop on missing or ambiguous
90
- row targets. Regenerate older tests and follow the [journal migration steps](docs/usage.md).
91
- See the [releases](https://github.com/chriswu727/argus/releases) and
92
- [changelog](CHANGELOG.md#0122---2026-11-08).
88
+ **0.13.0**: live verification, `verify_persistence` and exported tests now
89
+ decide with the same rules, checked against each other on every change. They
90
+ read what a user can see, in every frame and open shadow root, and an exported
91
+ test reports an inconclusive check instead of a false failure. Regenerate
92
+ exported tests to get the shared rules; see the
93
+ [journal migration steps](docs/usage.md#regression-gates-in-ci), the
94
+ [releases](https://github.com/chriswu727/argus/releases) and the
95
+ [changelog](CHANGELOG.md#0130---2026-11-12).
93
96
 
94
97
  ## Engineering
95
98
 
@@ -92,7 +92,6 @@ from argus.application.text import _short as _short
92
92
  from argus.application.text import _text_in_state as _text_in_state
93
93
  from argus.application.text import _token_present as _token_present
94
94
  from argus.application.text import _visible_text_in_state as _visible_text_in_state
95
- from argus.application.verification import _LOGOUT_PHRASES as _LOGOUT_PHRASES
96
95
  from argus.application.verification import _looks_logged_out as _looks_logged_out
97
96
  from argus.application.verification import _minimize_replay as _minimize_replay
98
97
  from argus.application.verification import _receipt_verdict as _receipt_verdict
@@ -100,3 +99,6 @@ from argus.application.verification import _run_replay_receipt as _run_replay_re
100
99
  from argus.application.verification import (
101
100
  _run_reproduction_check as _run_reproduction_check,
102
101
  )
102
+ from argus.domain.verdict import LOGIN_WALL_PHRASES
103
+
104
+ _LOGOUT_PHRASES = LOGIN_WALL_PHRASES
@@ -84,7 +84,10 @@ class AccessPolicy:
84
84
  return resolved
85
85
 
86
86
  def check_url(self, url: str) -> None:
87
- if url == "about:blank":
87
+ # Chromium's own error page (failed navigation) holds no site content;
88
+ # rejecting it would refuse every later call, including the navigate
89
+ # that leaves it.
90
+ if url == "about:blank" or url.startswith("chrome-error://"):
88
91
  return
89
92
  try:
90
93
  parsed = urlsplit(url)
@@ -17,6 +17,8 @@ class FileJournal:
17
17
  if any(
18
18
  not isinstance(entry.get("fingerprint"), str)
19
19
  or not isinstance(entry.get("verify"), dict)
20
+ or not isinstance(entry.get("title", ""), str)
21
+ or not isinstance(entry.get("severity", ""), str)
20
22
  for entry in entries
21
23
  ):
22
24
  raise StorageError(f"invalid finding entry in {path}")
@@ -61,11 +61,10 @@ class Monitor:
61
61
  )
62
62
  except ValueError:
63
63
  return False
64
- return (
65
- host in allowed
66
- and valid_origin
67
- and request.headers.get("sec-fetch-site") != "cross-site"
68
- )
64
+ # Another port on localhost is same-site, and a page served there (usually
65
+ # the app under test) must not be able to hold the event stream slots.
66
+ fetch_site = request.headers.get("sec-fetch-site")
67
+ return host in allowed and valid_origin and fetch_site in (None, "same-origin", "none")
69
68
 
70
69
  async def health(self, request: Request) -> Response:
71
70
  return JSONResponse({"status": "ok", "schema_version": 1})
@@ -2,6 +2,7 @@ from __future__ import annotations
2
2
 
3
3
  import base64
4
4
  import hashlib
5
+ import re
5
6
  from pathlib import Path
6
7
  from typing import List, Optional
7
8
  from urllib.parse import quote
@@ -14,6 +15,17 @@ from argus.domain.models import (
14
15
  Severity,
15
16
  )
16
17
 
18
+ _XML_ILLEGAL = re.compile("[^\t\n\r\u0020-\ud7ff\ue000-\ufffd\U00010000-\U0010ffff]")
19
+
20
+
21
+ def xml_safe(document: str) -> str:
22
+ """Drop characters XML 1.0 cannot carry (ANSI escapes, NUL, lone surrogates).
23
+
24
+ ElementTree serializes them as-is, which leaves a file no parser accepts.
25
+ """
26
+ return _XML_ILLEGAL.sub("", document)
27
+
28
+
17
29
  _BUGTYPE_LABELS = {
18
30
  BugType.CONSOLE_ERROR: "Console Error",
19
31
  BugType.NETWORK_ERROR: "Network Error",
@@ -78,10 +90,10 @@ def _repro_badge(receipt: Optional[dict]) -> str:
78
90
  "#b35900",
79
91
  f"NOT REPRODUCED · replayed {receipt.get('steps', '?')} steps, symptom absent",
80
92
  )
81
- elif receipt.get("expect_status") is not None:
82
- color, label = "#b35900", f"NOT REPRODUCED · expected HTTP {receipt['expect_status']}"
83
93
  elif receipt.get("flaky"):
84
94
  color, label = "#9a6700", f"INTERMITTENT · {receipt.get('runs', '')} on reload"
95
+ elif receipt.get("expect_status") is not None:
96
+ color, label = "#b35900", f"NOT REPRODUCED · expected HTTP {receipt['expect_status']}"
85
97
  else:
86
98
  color, label = (
87
99
  "#b35900",
@@ -22,6 +22,13 @@ from .common import (
22
22
  )
23
23
  from .privacy import safe_report
24
24
 
25
+ # A check with matches=None was inconclusive (login wall, HTTP error, redirect).
26
+ _CHECK_STYLE = {
27
+ True: ("cv-ok", "MATCH"),
28
+ False: ("cv-bad", "MISMATCH"),
29
+ None: ("cv-unknown", "INCONCLUSIVE"),
30
+ }
31
+
25
32
 
26
33
  @safe_report
27
34
  def render_html(
@@ -165,8 +172,8 @@ def render_html(
165
172
  reference_groups.append(f"<strong>Actions</strong><ul>{values}</ul>")
166
173
  if refs.get("verifications"):
167
174
  values = "".join(
168
- f'<li><span class="cv {"cv-ok" if check.get("matches") else "cv-bad"}">'
169
- f"{'MATCH' if check.get('matches') else 'MISMATCH'}</span> "
175
+ f'<li><span class="cv {_CHECK_STYLE[check.get("matches")][0]}">'
176
+ f"{_CHECK_STYLE[check.get('matches')][1]}</span> "
170
177
  f"{_esc(check.get('expect', ''))} {_esc(check.get('target_text', ''))} "
171
178
  f"@ <code>{_esc(check.get('url', ''))}</code></li>"
172
179
  for check in refs["verifications"]
@@ -307,6 +314,7 @@ h3{{font-size:1.1rem;margin:.5rem 0;color:#f1f5f9}}
307
314
  .cv{{display:inline-block;padding:1px 5px;border-radius:3px;font-size:.65rem;font-weight:700}}
308
315
  .cv-ok{{background:#166534;color:#dcfce7}}
309
316
  .cv-bad{{background:#9a3412;color:#ffedd5}}
317
+ .cv-unknown{{background:#475569;color:#f1f5f9}}
310
318
  .cs{{display:inline-block;padding:2px 7px;border-radius:4px;font-size:.7rem;font-weight:700;text-transform:uppercase;margin-right:.35rem}}
311
319
  .cs-exercised{{background:#166534;color:#dcfce7}}
312
320
  .cs-blocked{{background:#9a3412;color:#ffedd5}}
@@ -1,7 +1,9 @@
1
1
  from __future__ import annotations
2
2
 
3
3
  from argus.domain.models import ExplorationResult
4
+ from argus.domain.verification import receipt_outcome
4
5
 
6
+ from .common import xml_safe
5
7
  from .privacy import safe_report
6
8
 
7
9
 
@@ -14,9 +16,7 @@ def render_junit(r: ExplorationResult) -> str:
14
16
  import xml.etree.ElementTree as ET
15
17
 
16
18
  findings = [b.to_dict() for b in r.bugs]
17
- failures = sum(
18
- 1 for f in findings if (f.get("reproduction") or {}).get("reproduced") is not False
19
- )
19
+ failures = sum(1 for f in findings if receipt_outcome(f.get("reproduction")) is not False)
20
20
  suite = ET.Element(
21
21
  "testsuite",
22
22
  {
@@ -28,8 +28,16 @@ def render_junit(r: ExplorationResult) -> str:
28
28
  },
29
29
  )
30
30
  for f in findings:
31
- verdict = (f.get("reproduction") or {}).get("reproduced")
32
- tier = "PROVEN" if verdict is True else ("REFUTED" if verdict is False else "UNVERIFIED")
31
+ verdict = receipt_outcome(f.get("reproduction"))
32
+ tier = (
33
+ "PROVEN"
34
+ if verdict is True
35
+ else "REFUTED"
36
+ if verdict is False
37
+ else "INTERMITTENT"
38
+ if (f.get("reproduction") or {}).get("flaky")
39
+ else "UNVERIFIED"
40
+ )
33
41
  tc = ET.SubElement(
34
42
  suite,
35
43
  "testcase",
@@ -56,4 +64,6 @@ def render_junit(r: ExplorationResult) -> str:
56
64
  else:
57
65
  fail = ET.SubElement(tc, "failure", {"message": f["title"][:200], "type": f["type"]})
58
66
  fail.text = "\n".join(body_lines)
59
- return '<?xml version="1.0" encoding="UTF-8"?>\n' + ET.tostring(suite, encoding="unicode")
67
+ return '<?xml version="1.0" encoding="UTF-8"?>\n' + xml_safe(
68
+ ET.tostring(suite, encoding="unicode")
69
+ )
@@ -3,6 +3,7 @@ from __future__ import annotations
3
3
  from typing import Dict
4
4
 
5
5
  from argus.domain.models import ExplorationResult, Severity
6
+ from argus.domain.verification import receipt_outcome
6
7
 
7
8
  from .common import (
8
9
  _BUGTYPE_LABELS,
@@ -14,7 +15,13 @@ from .privacy import safe_report
14
15
  def render_sarif(r: ExplorationResult) -> str:
15
16
  import json
16
17
 
17
- _LEVEL = {Severity.HIGH: "error", Severity.MEDIUM: "warning", Severity.LOW: "note"}
18
+ _LEVEL = {
19
+ Severity.CRITICAL: "error",
20
+ Severity.HIGH: "error",
21
+ Severity.MEDIUM: "warning",
22
+ Severity.LOW: "note",
23
+ Severity.INFO: "note",
24
+ }
18
25
  rules: Dict[str, dict] = {}
19
26
  results = []
20
27
  for b in r.bugs:
@@ -26,7 +33,7 @@ def render_sarif(r: ExplorationResult) -> str:
26
33
  "name": rid,
27
34
  "shortDescription": {"text": _BUGTYPE_LABELS.get(b.type, rid)},
28
35
  }
29
- verdict = (d.get("reproduction") or {}).get("reproduced")
36
+ verdict = receipt_outcome(d.get("reproduction"))
30
37
  msg = d["title"] + (f" — {d['description']}" if d.get("description") else "")
31
38
  results.append(
32
39
  {
@@ -41,6 +48,7 @@ def render_sarif(r: ExplorationResult) -> str:
41
48
  "properties": {
42
49
  "verified": d["verified"],
43
50
  "reproduced": verdict,
51
+ "intermittent": bool((d.get("reproduction") or {}).get("flaky")),
44
52
  "severity": d["severity"],
45
53
  },
46
54
  }