@zhivex-ai/harness 1.0.0 → 1.1.0-rc.2

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 (181) hide show
  1. package/CHANGELOG.md +97 -1
  2. package/README.md +52 -306
  3. package/SECURITY.md +1 -1
  4. package/SUPPORT.md +7 -3
  5. package/benchmarks/README.md +3 -3
  6. package/contracts/client-protocol-v1.schema.json +341 -0
  7. package/contracts/public-api.json +100 -1
  8. package/contracts/stable-api-signatures.json +14 -14
  9. package/dist/approval-diff.d.ts +45 -0
  10. package/dist/approval-history.d.ts +38 -0
  11. package/dist/approval-preview.d.ts +26 -0
  12. package/dist/cli-credentials.d.ts +38 -0
  13. package/dist/cli-help.d.ts +3 -0
  14. package/dist/cli-options.d.ts +1 -1
  15. package/dist/cli-stream.d.ts +3 -3
  16. package/dist/cli.d.ts +15 -69
  17. package/dist/cli.js +13796 -10032
  18. package/dist/client-contract.d.ts +208 -0
  19. package/dist/console-commands.d.ts +6 -0
  20. package/dist/console-context.d.ts +5 -1
  21. package/dist/console-input.d.ts +44 -4
  22. package/dist/console-model-catalog.d.ts +25 -0
  23. package/dist/console-navigation.d.ts +27 -0
  24. package/dist/console-presentation.d.ts +18 -0
  25. package/dist/console-welcome.d.ts +13 -0
  26. package/dist/execution-environment.d.ts +4 -0
  27. package/dist/harness.d.ts +3 -0
  28. package/dist/hostile-repository-demo.js +1913 -798
  29. package/dist/index.d.ts +7 -0
  30. package/dist/index.js +7908 -6330
  31. package/dist/local-service.d.ts +34 -0
  32. package/dist/operations.d.ts +23 -1
  33. package/dist/orchestration.d.ts +3 -0
  34. package/dist/persistence-secrets.d.ts +2 -0
  35. package/dist/recorded-workspace.d.ts +4 -0
  36. package/dist/run-document.d.ts +91 -0
  37. package/dist/runtime-policy.d.ts +4 -4
  38. package/dist/service-cli.d.ts +2 -0
  39. package/dist/service-cli.js +10858 -0
  40. package/dist/service-client-cli.d.ts +2 -0
  41. package/dist/service-events.d.ts +50 -0
  42. package/dist/sessions.d.ts +3 -0
  43. package/dist/sqlite-access.d.ts +26 -0
  44. package/dist/sqlite-database.d.ts +2 -0
  45. package/dist/state-backup.d.ts +8 -0
  46. package/dist/task-memory.d.ts +2 -2
  47. package/dist/terminal-continuation.d.ts +3 -0
  48. package/dist/terminal-markdown.d.ts +1 -0
  49. package/dist/terminal-ui.d.ts +5 -0
  50. package/dist/usage-ledger.d.ts +84 -0
  51. package/dist/verify-historical-migrations.js +5699 -5191
  52. package/dist/workspace.d.ts +22 -0
  53. package/dist/zhx.js +13796 -10032
  54. package/docs/CLI.md +136 -30
  55. package/docs/CLIENT_PROTOCOL.md +212 -0
  56. package/docs/CREDENTIALS.md +59 -0
  57. package/docs/DURABLE_OPERATIONS.md +2 -2
  58. package/docs/FIRST_USE.md +82 -0
  59. package/docs/FIRST_USE_EXAMPLE.md +40 -0
  60. package/docs/HOSTILE_REPOSITORY_DEMO.md +1 -1
  61. package/docs/LOCAL_SERVICE.md +142 -0
  62. package/docs/README.md +24 -23
  63. package/docs/THREAT_MODEL.md +1 -1
  64. package/docs/TIME_TO_SAFE_FIX.md +2 -2
  65. package/docs/USAGE.md +181 -0
  66. package/evaluations/daily-cli-cohort.json +33 -0
  67. package/evaluations/representative-assembly-matrix.json +19 -1
  68. package/examples/README.md +26 -0
  69. package/examples/first-use/greeting.test.ts +3 -0
  70. package/examples/first-use/greeting.ts +1 -0
  71. package/examples/first-use/package.json +7 -0
  72. package/package.json +34 -9
  73. package/ROADMAP.md +0 -414
  74. package/benchmarks/baselines/context-navigation-2026-09-09.json +0 -94
  75. package/benchmarks/baselines/efficiency-2026-09-08.json +0 -233
  76. package/benchmarks/baselines/external-final-2026-09-08.json +0 -588
  77. package/benchmarks/baselines/external-intermediate-2026-09-08.json +0 -116
  78. package/benchmarks/baselines/external-matched-openai-2026-09-08.json +0 -1922
  79. package/benchmarks/baselines/external-matched-qwen-2026-09-08.json +0 -1706
  80. package/benchmarks/baselines/external-pilot-2026-09-08.json +0 -439
  81. package/benchmarks/baselines/external-qwen-flash-2026-09-08.json +0 -1570
  82. package/benchmarks/baselines/external-qwen-remediation-2026-09-09.json +0 -1509
  83. package/benchmarks/baselines/external-qwen-sdk-114-progress-2026-09-09.json +0 -2509
  84. package/benchmarks/baselines/external-sdk-upgrade-2026-09-08.json +0 -564
  85. package/benchmarks/baselines/external-verifier-recovery-2026-09-09.json +0 -794
  86. package/benchmarks/baselines/full-harness-review-2026-09-10.json +0 -134
  87. package/benchmarks/baselines/harness-audit-2026-09-09.json +0 -136
  88. package/benchmarks/baselines/harness-remediation-2026-09-09.json +0 -160
  89. package/benchmarks/baselines/oci-remediation-2026-09-10.json +0 -535
  90. package/benchmarks/baselines/rc14-holdout-2026-09-10.json +0 -854
  91. package/benchmarks/baselines/repair-architecture-audit-2026-09-09.json +0 -75
  92. package/benchmarks/baselines/time-to-safe-fix-live-expanded-2026-08-21.json +0 -186
  93. package/benchmarks/baselines/time-to-safe-fix-live-smoke-2026-08-21.json +0 -186
  94. package/docs/GA_READINESS.md +0 -29
  95. package/docs/GEMINI_1_0_DECISION.md +0 -9
  96. package/docs/LIVE_CERTIFICATION.md +0 -107
  97. package/docs/RC13_SECURITY_REVIEW.md +0 -28
  98. package/docs/RC14_DELEGATED_SECURITY_DECISION.md +0 -80
  99. package/docs/RC14_SECURITY_REVIEW.md +0 -40
  100. package/docs/RELEASE.md +0 -96
  101. package/docs/ROLLBACK.md +0 -9
  102. package/docs/SECURITY_REVIEW_EVIDENCE.md +0 -17
  103. package/docs/ga-readiness.json +0 -366
  104. package/docs/reports/EFFICIENCY_REVIEW.md +0 -156
  105. package/docs/reports/EXTERNAL_COMPARISON.md +0 -434
  106. package/docs/reports/FULL_HARNESS_REVIEW_2026-09-10.md +0 -200
  107. package/docs/reports/GA_BUDGET_REMEDIATION_2026-09-19.md +0 -38
  108. package/docs/reports/GA_REPAIR_RECOVERY_2026-09-19.md +0 -700
  109. package/docs/reports/GA_VALIDATION_2026-09-19.md +0 -111
  110. package/docs/reports/HARNESS_ARCHITECTURE_REVIEW_2026-09-09.md +0 -202
  111. package/docs/reports/HARNESS_AUDIT_2026-09-09.md +0 -306
  112. package/docs/reports/HARNESS_REMEDIATION_2026-09-09.md +0 -74
  113. package/docs/reports/HARNESS_REMEDIATION_2026-09-10.md +0 -90
  114. package/docs/reports/MODEL_COMPARISON_2026-09-09.md +0 -184
  115. package/docs/reports/QWEN_REMEDIATION_COMPARISON_2026-09-09.md +0 -142
  116. package/docs/reports/RC14_VALIDATION_2026-09-10.md +0 -45
  117. package/docs/reports/README.md +0 -26
  118. package/docs/reports/evidence/ga-budget-remediation-2026-09-19.json +0 -133
  119. package/docs/reports/evidence/ga-catalog-ablation-2026-09-19.json +0 -699
  120. package/docs/reports/evidence/ga-catalog-ablation-plan-2026-09-19.json +0 -40
  121. package/docs/reports/evidence/ga-catalog-observed-2026-09-19.json +0 -985
  122. package/docs/reports/evidence/ga-catalog-observed-plan-2026-09-19.json +0 -28
  123. package/docs/reports/evidence/ga-catalog-observed-status-2026-09-19.json +0 -15
  124. package/docs/reports/evidence/ga-complete-local-check-2026-09-19.json +0 -47
  125. package/docs/reports/evidence/ga-completion-reminder-2026-09-19.json +0 -22
  126. package/docs/reports/evidence/ga-fresh-cohort-plan-2026-09-19.json +0 -54
  127. package/docs/reports/evidence/ga-fresh-cohort-preparation-2026-09-19.json +0 -242
  128. package/docs/reports/evidence/ga-fresh-cohort-result-2026-09-19.json +0 -4318
  129. package/docs/reports/evidence/ga-local-validation-followup-2026-09-19.json +0 -66
  130. package/docs/reports/evidence/ga-node-consumers-2026-09-19.json +0 -31
  131. package/docs/reports/evidence/ga-package-content-validation-2026-09-19.json +0 -22
  132. package/docs/reports/evidence/ga-plan-completion-2026-09-19.json +0 -39
  133. package/docs/reports/evidence/ga-plan-reserve-2026-09-19.json +0 -573
  134. package/docs/reports/evidence/ga-plan-reserve-plan-2026-09-19.json +0 -24
  135. package/docs/reports/evidence/ga-plan-transition-2026-09-19.json +0 -19
  136. package/docs/reports/evidence/ga-private-verifier-plan-2026-09-19.json +0 -24
  137. package/docs/reports/evidence/ga-private-verifier-root-cause-2026-09-19.json +0 -41
  138. package/docs/reports/evidence/ga-published-sdk-2026-09-19.json +0 -35
  139. package/docs/reports/evidence/ga-python-checkout-imports-2026-09-19.json +0 -95
  140. package/docs/reports/evidence/ga-python-origin-delivery-2026-09-19.json +0 -211
  141. package/docs/reports/evidence/ga-python-origin-delivery-plan-2026-09-19.json +0 -27
  142. package/docs/reports/evidence/ga-qwen-patched-package-2026-09-19.json +0 -1092
  143. package/docs/reports/evidence/ga-qwen-patched-package-plan-2026-09-19.json +0 -18
  144. package/docs/reports/evidence/ga-qwen-published-2026-09-19.json +0 -571
  145. package/docs/reports/evidence/ga-qwen-published-plan-2026-09-19.json +0 -18
  146. package/docs/reports/evidence/ga-qwen-tool-choice-2026-09-19.json +0 -85
  147. package/docs/reports/evidence/ga-qwen-tool-choice-patched-2026-09-19.json +0 -85
  148. package/docs/reports/evidence/ga-recovery-development-2026-09-19.json +0 -1066
  149. package/docs/reports/evidence/ga-recovery-graded-development-2026-09-19.json +0 -1313
  150. package/docs/reports/evidence/ga-recovery-graded-v2-2026-09-19.json +0 -1151
  151. package/docs/reports/evidence/ga-recovery-v3-2026-09-19.json +0 -3943
  152. package/docs/reports/evidence/ga-recovery-v4-2026-09-19.json +0 -1055
  153. package/docs/reports/evidence/ga-recovery-v5-2026-09-19.json +0 -1245
  154. package/docs/reports/evidence/ga-recovery-v6-2026-09-19.json +0 -699
  155. package/docs/reports/evidence/ga-recovery-v7-2026-09-19.json +0 -854
  156. package/docs/reports/evidence/ga-recovery-v8-2026-09-19.json +0 -596
  157. package/docs/reports/evidence/ga-required-delivery-2026-09-19.json +0 -25
  158. package/docs/reports/evidence/ga-required-delivery-live-2026-09-19.json +0 -1337
  159. package/docs/reports/evidence/ga-required-delivery-live-plan-2026-09-19.json +0 -25
  160. package/docs/reports/evidence/ga-retained-candidate-grading-2026-09-19.json +0 -30
  161. package/docs/reports/evidence/ga-tool-selection-live-2026-09-19.json +0 -559
  162. package/docs/reports/evidence/ga-tool-selection-live-plan-2026-09-19.json +0 -28
  163. package/docs/reports/evidence/ga-tool-selection-recovery-2026-09-19.json +0 -27
  164. package/docs/reports/evidence/ga-validation-2026-09-19.json +0 -236
  165. package/docs/reports/evidence/ga-verifier-discovery-live-2026-09-19.json +0 -1369
  166. package/docs/reports/evidence/ga-verifier-discovery-plan-2026-09-19.json +0 -29
  167. package/docs/reports/evidence/ga-verifier-failure-identification-2026-09-19.json +0 -30
  168. package/docs/reports/evidence/ga-verifier-feedback-2026-09-19.json +0 -65
  169. package/docs/reports/evidence/ga-verifier-hints-live-2026-09-19.json +0 -1297
  170. package/docs/reports/evidence/ga-verifier-hints-plan-2026-09-19.json +0 -27
  171. package/docs/reports/evidence/ga-work-boundary-context-analysis-2026-09-19.json +0 -113
  172. package/docs/reports/evidence/ga-work-boundary-live-2026-09-19.json +0 -301
  173. package/docs/reports/evidence/ga-work-boundary-live-plan-2026-09-19.json +0 -84
  174. package/docs/reports/evidence/ga-work-budget-planning-2026-09-19.json +0 -32
  175. package/docs/reports/evidence/ga-working-context-2026-09-19.json +0 -22
  176. package/docs/reports/evidence/rc14-delegated-security-review-2026-09-20.json +0 -81
  177. package/docs/reports/evidence/rc14-ga-gate-2026-09-20.json +0 -18
  178. package/docs/reports/evidence/rc14-representative-2026-09-20.json +0 -509
  179. package/docs/reports/evidence/rc14-security-regressions-2026-09-20.json +0 -64
  180. package/docs/reports/evidence/security-current-2026-09-19.json +0 -91
  181. package/docs/reports/evidence/security-remediation-2026-09-19.json +0 -97
@@ -1,794 +0,0 @@
1
- {
2
- "schemaVersion": 1,
3
- "dataset": "princeton-nlp/SWE-bench_Verified",
4
- "revision": "c104f840cc67f8b6eec6f759ebc8b2693d585d4a",
5
- "model": "gpt-5.6-luna",
6
- "provider": "openai",
7
- "reasoning": "low",
8
- "transport": "openai-responses",
9
- "limits": {
10
- "steps": 24,
11
- "outputPerTurn": 2048,
12
- "inputTokens": 100000,
13
- "outputTokens": 16000,
14
- "timeoutSeconds": 300,
15
- "memoryMb": 2048,
16
- "cpus": 2
17
- },
18
- "versions": {
19
- "mini-swe-agent": "2.4.6",
20
- "swebench": "4.1.0"
21
- },
22
- "implementation": {
23
- "gitCommit": "0a7b69dddaa70a54b6b5b4097691c8112d4d76eb",
24
- "sources": {
25
- "bun.lock": "d5436a51773d94333612c611a162a3e63765fec88d6bbde2a71879662b49b5cd",
26
- "evaluations/external/requirements.lock": "42e90a08f10f821ebb958af79256f707a9fb5cf365277c4f9e2715d612028179",
27
- "package.json": "ade3c3a4c0f8ee193c560637bf96f1fe2782f5bc06a42dcdbb320bd7a932a173",
28
- "scripts/benchmark-swebench.ts": "ed17063e1847017ba2e72f14732b6ff8f5b1af7cd5253b91302af75f2a9e8476",
29
- "scripts/swebench/common.py": "c19cfc30b85dc3ad8dbf1ba6a4165e72ce96f624eb1721b613cfa3207636f7be",
30
- "scripts/swebench/mini_driver.py": "bd0a77221878370ec44760122a51f07fb56d969612afa9a7f171248b2958baa6",
31
- "scripts/swebench/model-budget.ts": "67b74d45d34235586773b7cf0db4d320306b14a6f432d53c084f80563ccdfa44",
32
- "scripts/swebench/preflight.py": "aed092a75dd52f504fba6d14814b69bd627ff17d0e311a1f1279f8b613b69675",
33
- "scripts/swebench/prepare.py": "bb7edd3fba5dbbe26597c97d744c60b94dbb7bcdfd7740a4a058c8583bd27e4c",
34
- "scripts/swebench/report.py": "70afcdf0d1f081014bb75d878f3ddf9195959e96b523ddf1e0ad5d3ec8273549",
35
- "scripts/swebench/run.py": "b184258e7c32313c9c801b50d28d1a9419d0bd2b077a422012d8fc6bc449c916",
36
- "scripts/swebench/telemetry.ts": "4c6c0d58641b91539d945f341b368a9c360b6ef89ff3abb9d2f33d3cb133c12e",
37
- "scripts/swebench/test_contracts.py": "a8c1348f1dff2e563c139e6d3e83eb141b4f2a71f91a58e2c5c051ed2a431024",
38
- "scripts/swebench/zhivex-driver.ts": "a538c1a83e3c4c0b93ab52c5e4308a6c387fbb5d45499708e12aa27cc25847ec",
39
- "scripts/time-to-safe-fix-efficiency.ts": "dd20fd8839280332ae4ca80d6846e98e07ff15d79e200b206fa2ba6be2724d2f",
40
- "src/bounded-reads.ts": "9807bb460894be3279c5425af19863d1c6e419a1582ba6910ec7e7fc35703ed6",
41
- "src/capabilities.ts": "56fdb60a2f6e1d33b2dc26d38b6d2709807e491ba1268d47b97876c2c0591d65",
42
- "src/change-envelope.ts": "9047ec978c9e5117965b8093ec961d855f97bbc2c3accdc79b74cfe6a0e4f01e",
43
- "src/cli-options.ts": "97781386289e3877c1318e4c94698ca7aaf2f256b89534dfcb21e966eb7d939f",
44
- "src/cli-profiles.ts": "4fe451f1645e95f4957597a8724fa989b3c661ae1693252bfd6e74a51d20fc74",
45
- "src/cli-stream.ts": "5ab8a4b1b8cf0a3dfa7b5b1f9e4e0facf8f6e7f6c341bdf700a41dc8927c825e",
46
- "src/cli.ts": "cce6f58dfbc543acdc029b188d4394cd059d940bdd361351fd877ae2d6e94880",
47
- "src/compaction.ts": "0c4f95d2f49c5154c9b87b3f79feca18227e274a8aacd2445e387a857a37853f",
48
- "src/config.ts": "c160faaf2d9fb637c5a08a7aff65d761a12c707ed23ebb87ab0e8519e9536bf3",
49
- "src/console-context.ts": "0ad7e4559504b9ed09ab500d615dea09af1075d357f02ec3b77f003d55d951db",
50
- "src/console-input.ts": "dc8ea0cdbba1364db288d5156c5ff9819362c46b37ef35a98b0475b6c445b1f8",
51
- "src/context-engineering.ts": "1995e564f6b91974e75bf66dd421ddcbea29b1f8eee59b76eb4e34c9d6d8f47e",
52
- "src/edit-contracts.ts": "47cc3132e27f8749d54a32bbcc33bb174cf4448f0d0171b2578cdb559cf2176b",
53
- "src/errors.ts": "e02f4eb80562a83762f908bda56c950ec6a2b78725c0c8a49f41fee0057ab903",
54
- "src/execution-environment.ts": "8519e532dfafe4ae51bfbc81e0732267bc6aa9eec871ffde8c917aa001e358e5",
55
- "src/file-security.ts": "4590b46c8e4b9d3891a75cd0d0ada14e0ae7b58e8e4bdf82f68df688a1fbb750",
56
- "src/harness.ts": "235fa587e89f39f204c40e1031681c477f3e5db0cfc8d9e1c3325575700fa26b",
57
- "src/index.ts": "f791311f14dc27105191d97bb5454748d10beab827da9aba3ea949ffc32fce7c",
58
- "src/json-contracts.ts": "0702e5b0ff16ff6304c74b4101bd31a3fa8eabdebe267d312b56d3871a7e302c",
59
- "src/mcp.ts": "5f3f7a35b89063264ae18ab9273b84e2e59ac8c8e611a6b64be377ba61cc1ac5",
60
- "src/operations.ts": "0775da79208c6f5d18476cc76404e84ddda610979d9d0e5dc15035aa6999613c",
61
- "src/orchestration.ts": "7383da1047363819e6560f51892fd4082e5337856a3d889cdc076f021273774e",
62
- "src/package-manager.ts": "cec4666bbab2f42e0a269709631f79f9f87c2c4dbb3cafeffacfaff761069e2c",
63
- "src/process-runtime.ts": "b7286731a0b5c96dcad0b8e3a4bf1155d0539d73303cd7ebe77ffabf858f1729",
64
- "src/providers.ts": "70ccfc7f315cd7238e83d69b719ce4908f3b620c564cbb4f00f93e102e025602",
65
- "src/replacement-edits.ts": "8a0efdd3a7a9fc3e5a5475916cbd17ef882265881bb61076db58673e80b72957",
66
- "src/routing.ts": "8e7c9f5968b4bd638cbbc86b19847ca183f3418b9a298783b96a88cffa568e22",
67
- "src/run-interruption.ts": "df99129cf2dd6b938eca629cb897c4350a7d57fe6bf0ce437f5f3f30271617df",
68
- "src/sessions.ts": "5167c8cf36bd4b8845ee278883394700321b65397f49c1dda0cf0d0de09dd06e",
69
- "src/sqlite-database.ts": "18765d1ca0c2f2ad87bfd82d311312dfe8a42006556f27bc21da5892ed3b044e",
70
- "src/state-backup.ts": "b6f8f1aedeb5dce31b65c23c7ffc50030870655b58a5aaaef9989bafb9da5c68",
71
- "src/state-directory.ts": "3711550991951e6b738a08deaa0e06da84d01864c03a9602600f592963fa3211",
72
- "src/terminal-markdown.ts": "8cffb6ebaf25108cb2bee04956d9b2973635b010f5d7ce6be21b3ceffe3e8bfd",
73
- "src/terminal-ui.ts": "e50025a8649d1f60efdf3af895b8639730ec2e86553fb80a61a776b11076c705",
74
- "src/time-to-safe-fix.ts": "4add1d49c0ea209cc496c66fa2b2012dc7bcab87e678a58c8591b3c2cdc8df8a",
75
- "src/version.ts": "b2c2f80a40adc08e8258b81e82434273f0bdfa787528b2221f14efe7f06aa747",
76
- "src/workspace.ts": "e9c2ed5c950d751d0f93a5512584d0698449b551d3270b3e115ccca380565222"
77
- }
78
- },
79
- "host": {
80
- "system": "Darwin",
81
- "architecture": "arm64",
82
- "python": "3.12.13",
83
- "docker": "29.7.2"
84
- },
85
- "approvalPolicy": "Automated operator approves supported isolated Zhivex tools; successful approved verification and patch import is a terminal receipt. Human wait is zero. Not a human-UX comparison.",
86
- "taskIds": [
87
- "scikit-learn__scikit-learn-14141",
88
- "django__django-11265"
89
- ],
90
- "repetitions": 1,
91
- "report": {
92
- "schemaVersion": 1,
93
- "kind": "swebench-comparison",
94
- "manifestSha256": "bfb84c6d86b015dfb21f7fd2f3ba152687122932b5e10eec513e9ac12b414ab5",
95
- "complete": true,
96
- "candidates": {
97
- "zhivex": {
98
- "planned": 2,
99
- "recorded": 2,
100
- "resolved": 1,
101
- "resolutionRate": 0.5,
102
- "missing": 0,
103
- "gradingFailures": 0,
104
- "usageComplete": true,
105
- "totalCostUsd": null,
106
- "costPerResolvedUsd": null,
107
- "resolvedP50Ms": 61794.38716697041
108
- },
109
- "mini-swe-agent": {
110
- "planned": 2,
111
- "recorded": 2,
112
- "resolved": 1,
113
- "resolutionRate": 0.5,
114
- "missing": 0,
115
- "gradingFailures": 0,
116
- "usageComplete": true,
117
- "totalCostUsd": null,
118
- "costPerResolvedUsd": null,
119
- "resolvedP50Ms": 47283.00745802699
120
- }
121
- },
122
- "pairedTaskResolutionDifference": 0.0,
123
- "pairedTaskBootstrap95": [
124
- 0.0,
125
- 0.0
126
- ],
127
- "evidenceBoundary": "Local matched-model comparison. Missing and failed runs remain in the denominator. Small-sample intervals are exploratory. No safety superiority is inferred from SWE-bench correctness."
128
- },
129
- "samples": [
130
- {
131
- "candidate": "zhivex",
132
- "instanceId": "scikit-learn__scikit-learn-14141",
133
- "repetition": 0,
134
- "resolved": true,
135
- "gradingStatus": "completed",
136
- "usageComplete": true,
137
- "costUsd": null,
138
- "schemaVersion": 1,
139
- "status": "completed",
140
- "failure": null,
141
- "phase": "agent",
142
- "durationMs": 37294.750374999996,
143
- "approvals": 7,
144
- "turns": [
145
- {
146
- "index": 1,
147
- "status": "completed",
148
- "inputTokens": 1969,
149
- "outputTokens": 113,
150
- "cachedInputTokens": 1966,
151
- "requestCharacters": 4396,
152
- "tools": [
153
- {
154
- "name": "list_files",
155
- "isError": false,
156
- "characters": 8308
157
- },
158
- {
159
- "name": "search_many",
160
- "isError": false,
161
- "characters": 4135
162
- }
163
- ]
164
- },
165
- {
166
- "index": 2,
167
- "status": "completed",
168
- "inputTokens": 5843,
169
- "outputTokens": 174,
170
- "cachedInputTokens": 1966,
171
- "requestCharacters": 17045,
172
- "tools": [
173
- {
174
- "name": "read_files",
175
- "isError": false,
176
- "characters": 5641
177
- },
178
- {
179
- "name": "search_many",
180
- "isError": false,
181
- "characters": 2334
182
- }
183
- ]
184
- },
185
- {
186
- "index": 3,
187
- "status": "waiting_approval",
188
- "inputTokens": 8679,
189
- "outputTokens": 125,
190
- "cachedInputTokens": 5840,
191
- "requestCharacters": 13081,
192
- "tools": []
193
- },
194
- {
195
- "index": 4,
196
- "status": "completed",
197
- "inputTokens": 2721,
198
- "outputTokens": 77,
199
- "cachedInputTokens": 1876,
200
- "requestCharacters": 11693,
201
- "tools": [
202
- {
203
- "name": "run_environment_command",
204
- "isError": false,
205
- "characters": 1827,
206
- "exitCode": 1,
207
- "timedOut": false
208
- }
209
- ]
210
- },
211
- {
212
- "index": 5,
213
- "status": "waiting_approval",
214
- "inputTokens": 4342,
215
- "outputTokens": 186,
216
- "cachedInputTokens": 2718,
217
- "requestCharacters": 8666,
218
- "tools": [
219
- {
220
- "name": "read_files",
221
- "isError": false,
222
- "characters": 4644
223
- }
224
- ]
225
- },
226
- {
227
- "index": 6,
228
- "status": "completed",
229
- "inputTokens": 4744,
230
- "outputTokens": 27,
231
- "cachedInputTokens": 4339,
232
- "requestCharacters": 26101,
233
- "tools": [
234
- {
235
- "name": "apply_reviewed_replacement",
236
- "isError": false,
237
- "characters": 475
238
- }
239
- ]
240
- },
241
- {
242
- "index": 7,
243
- "status": "completed",
244
- "inputTokens": 4986,
245
- "outputTokens": 333,
246
- "cachedInputTokens": 4741,
247
- "requestCharacters": 4098,
248
- "tools": [
249
- {
250
- "name": "inspect_environment_patch",
251
- "isError": false,
252
- "characters": 499
253
- },
254
- {
255
- "name": "verify_and_apply_environment_patch",
256
- "isError": true,
257
- "characters": 276,
258
- "exitCode": 1,
259
- "timedOut": false,
260
- "diagnostic": {
261
- "code": "EXECUTION_FAILED",
262
- "category": "execution",
263
- "retryable": false,
264
- "chain": [
265
- {
266
- "providerReason": null,
267
- "validationIssues": [],
268
- "status": null,
269
- "errorClass": "Object",
270
- "providerHints": [],
271
- "fingerprint": "d22955ecb1810cd7e960e732636eec570e04ab3b19eab276ebf23e2096a3c941",
272
- "reason": null
273
- }
274
- ]
275
- }
276
- }
277
- ]
278
- },
279
- {
280
- "index": 8,
281
- "status": "waiting_approval",
282
- "inputTokens": 5354,
283
- "outputTokens": 137,
284
- "cachedInputTokens": 4983,
285
- "requestCharacters": 36680,
286
- "tools": []
287
- },
288
- {
289
- "index": 9,
290
- "status": "completed",
291
- "inputTokens": 3232,
292
- "outputTokens": 39,
293
- "cachedInputTokens": 1876,
294
- "requestCharacters": 13206,
295
- "tools": [
296
- {
297
- "name": "run_environment_command",
298
- "isError": false,
299
- "characters": 2823,
300
- "exitCode": 0,
301
- "timedOut": false
302
- }
303
- ]
304
- },
305
- {
306
- "index": 10,
307
- "status": "waiting_approval",
308
- "inputTokens": 3486,
309
- "outputTokens": 57,
310
- "cachedInputTokens": 3229,
311
- "requestCharacters": 4314,
312
- "tools": [
313
- {
314
- "name": "inspect_environment_patch",
315
- "isError": false,
316
- "characters": 499
317
- }
318
- ]
319
- },
320
- {
321
- "index": 11,
322
- "status": "waiting_approval",
323
- "inputTokens": 5246,
324
- "outputTokens": 83,
325
- "cachedInputTokens": 3483,
326
- "requestCharacters": 29152,
327
- "tools": [
328
- {
329
- "name": "run_environment_command",
330
- "isError": false,
331
- "characters": 7713,
332
- "exitCode": 129,
333
- "timedOut": false
334
- }
335
- ]
336
- },
337
- {
338
- "index": 12,
339
- "status": "completed",
340
- "inputTokens": 5769,
341
- "outputTokens": 296,
342
- "cachedInputTokens": 5243,
343
- "requestCharacters": 35250,
344
- "tools": [
345
- {
346
- "name": "run_environment_command",
347
- "isError": false,
348
- "characters": 1469,
349
- "exitCode": 4,
350
- "timedOut": false
351
- },
352
- {
353
- "name": "verify_and_apply_environment_patch",
354
- "isError": false,
355
- "characters": 1469,
356
- "exitCode": 0,
357
- "timedOut": false
358
- }
359
- ]
360
- }
361
- ],
362
- "inputTokens": 56371,
363
- "outputTokens": 1647,
364
- "cachedInputTokens": 42260,
365
- "modelCalls": 12,
366
- "compactions": [
367
- {
368
- "beforeStep": 4,
369
- "estimatedTokensBefore": 10285,
370
- "estimatedTokensAfter": 2924
371
- },
372
- {
373
- "beforeStep": 9,
374
- "estimatedTokensBefore": 11043,
375
- "estimatedTokensAfter": 3302
376
- }
377
- ],
378
- "terminalDiagnostic": null,
379
- "thrownDiagnostic": null,
380
- "guardrails": [],
381
- "toolCalls": 14,
382
- "patchSha256": "87492f68fca7dbd335efa8c185dd242dfd3e4017089fab9dfe789a919959b5cf",
383
- "protectedFilesChanged": [],
384
- "officialResolved": true,
385
- "gradingDurationMs": 21972.350791969802,
386
- "totalDurationMs": 61794.38716697041
387
- },
388
- {
389
- "candidate": "mini-swe-agent",
390
- "instanceId": "scikit-learn__scikit-learn-14141",
391
- "repetition": 0,
392
- "resolved": true,
393
- "gradingStatus": "completed",
394
- "usageComplete": true,
395
- "costUsd": null,
396
- "schemaVersion": 1,
397
- "status": "Submitted",
398
- "failure": null,
399
- "phase": "agent",
400
- "durationMs": 25955.564584000967,
401
- "approvals": 0,
402
- "toolCalls": 9,
403
- "inputTokens": 38342,
404
- "outputTokens": 1049,
405
- "cachedInputTokens": 32148,
406
- "modelCalls": 7,
407
- "patchSha256": "b1ca3d45bf4e8d854c1921ac8f47ff519d0987ef33adc3c787e2a2c68bb2aba4",
408
- "protectedFilesChanged": [],
409
- "officialResolved": true,
410
- "gradingDurationMs": 19747.161707957275,
411
- "totalDurationMs": 47283.00745802699
412
- },
413
- {
414
- "candidate": "mini-swe-agent",
415
- "instanceId": "django__django-11265",
416
- "repetition": 0,
417
- "resolved": false,
418
- "gradingStatus": "completed",
419
- "usageComplete": true,
420
- "costUsd": null,
421
- "schemaVersion": 1,
422
- "status": "Submitted",
423
- "failure": null,
424
- "phase": "agent",
425
- "durationMs": 33420.02133303322,
426
- "approvals": 0,
427
- "toolCalls": 9,
428
- "inputTokens": 65632,
429
- "outputTokens": 1508,
430
- "cachedInputTokens": 57960,
431
- "modelCalls": 9,
432
- "patchSha256": "8d9bea95aa218e4dce77ebbe2227e11c3114110a04aa5517a4f9150c2780043d",
433
- "protectedFilesChanged": [],
434
- "officialResolved": false,
435
- "gradingDurationMs": 25020.768625021446,
436
- "totalDurationMs": 59974.00999994716
437
- },
438
- {
439
- "candidate": "zhivex",
440
- "instanceId": "django__django-11265",
441
- "repetition": 0,
442
- "resolved": false,
443
- "gradingStatus": "completed",
444
- "usageComplete": true,
445
- "costUsd": null,
446
- "schemaVersion": 1,
447
- "status": "failed",
448
- "failure": "EXECUTION_FAILED",
449
- "phase": "agent",
450
- "durationMs": 32650.040167,
451
- "approvals": 0,
452
- "turns": [
453
- {
454
- "index": 1,
455
- "status": "completed",
456
- "inputTokens": 3015,
457
- "outputTokens": 42,
458
- "cachedInputTokens": 3012,
459
- "requestCharacters": 2,
460
- "tools": [
461
- {
462
- "name": "list_files",
463
- "isError": false,
464
- "characters": 9866
465
- }
466
- ]
467
- },
468
- {
469
- "index": 2,
470
- "status": "completed",
471
- "inputTokens": 5778,
472
- "outputTokens": 117,
473
- "cachedInputTokens": 3012,
474
- "requestCharacters": 2,
475
- "tools": [
476
- {
477
- "name": "search_many",
478
- "isError": false,
479
- "characters": 565
480
- },
481
- {
482
- "name": "search_many",
483
- "isError": false,
484
- "characters": 5001
485
- }
486
- ]
487
- },
488
- {
489
- "index": 3,
490
- "status": "completed",
491
- "inputTokens": 7766,
492
- "outputTokens": 109,
493
- "cachedInputTokens": 5775,
494
- "requestCharacters": 2,
495
- "tools": [
496
- {
497
- "name": "read_files",
498
- "isError": false,
499
- "characters": 10504
500
- },
501
- {
502
- "name": "read_files",
503
- "isError": false,
504
- "characters": 1755
505
- }
506
- ]
507
- },
508
- {
509
- "index": 4,
510
- "status": "completed",
511
- "inputTokens": 5511,
512
- "outputTokens": 151,
513
- "cachedInputTokens": 1876,
514
- "requestCharacters": 19220,
515
- "tools": [
516
- {
517
- "name": "read_files",
518
- "isError": false,
519
- "characters": 10920
520
- },
521
- {
522
- "name": "search_many",
523
- "isError": false,
524
- "characters": 3876
525
- }
526
- ]
527
- },
528
- {
529
- "index": 5,
530
- "status": "completed",
531
- "inputTokens": 9840,
532
- "outputTokens": 79,
533
- "cachedInputTokens": 5508,
534
- "requestCharacters": 2,
535
- "tools": [
536
- {
537
- "name": "read_files",
538
- "isError": false,
539
- "characters": 13267
540
- }
541
- ]
542
- },
543
- {
544
- "index": 6,
545
- "status": "completed",
546
- "inputTokens": 5635,
547
- "outputTokens": 89,
548
- "cachedInputTokens": 1876,
549
- "requestCharacters": 20066,
550
- "tools": [
551
- {
552
- "name": "search_many",
553
- "isError": false,
554
- "characters": 5824
555
- }
556
- ]
557
- },
558
- {
559
- "index": 7,
560
- "status": "completed",
561
- "inputTokens": 7538,
562
- "outputTokens": 98,
563
- "cachedInputTokens": 5632,
564
- "requestCharacters": 2,
565
- "tools": [
566
- {
567
- "name": "read_files",
568
- "isError": false,
569
- "characters": 18981
570
- }
571
- ]
572
- },
573
- {
574
- "index": 8,
575
- "status": "completed",
576
- "inputTokens": 7391,
577
- "outputTokens": 163,
578
- "cachedInputTokens": 1876,
579
- "requestCharacters": 25956,
580
- "tools": [
581
- {
582
- "name": "search_many",
583
- "isError": false,
584
- "characters": 3412
585
- },
586
- {
587
- "name": "search_many",
588
- "isError": false,
589
- "characters": 361
590
- }
591
- ]
592
- },
593
- {
594
- "index": 9,
595
- "status": "completed",
596
- "inputTokens": 8812,
597
- "outputTokens": 79,
598
- "cachedInputTokens": 7388,
599
- "requestCharacters": 332,
600
- "tools": [
601
- {
602
- "name": "read_files",
603
- "isError": false,
604
- "characters": 12016
605
- }
606
- ]
607
- },
608
- {
609
- "index": 10,
610
- "status": "completed",
611
- "inputTokens": 5409,
612
- "outputTokens": 86,
613
- "cachedInputTokens": 1876,
614
- "requestCharacters": 332,
615
- "tools": [
616
- {
617
- "name": "search_many",
618
- "isError": false,
619
- "characters": 1094
620
- }
621
- ]
622
- },
623
- {
624
- "index": 11,
625
- "status": "completed",
626
- "inputTokens": 5846,
627
- "outputTokens": 58,
628
- "cachedInputTokens": 1876,
629
- "requestCharacters": 24434,
630
- "tools": [
631
- {
632
- "name": "read_files",
633
- "isError": false,
634
- "characters": 10628
635
- }
636
- ]
637
- },
638
- {
639
- "index": 12,
640
- "status": "completed",
641
- "inputTokens": 8818,
642
- "outputTokens": 134,
643
- "cachedInputTokens": 1876,
644
- "requestCharacters": 36081,
645
- "tools": [
646
- {
647
- "name": "read_files",
648
- "isError": false,
649
- "characters": 9889
650
- }
651
- ]
652
- },
653
- {
654
- "index": 13,
655
- "status": "completed",
656
- "inputTokens": 5101,
657
- "outputTokens": 146,
658
- "cachedInputTokens": 1876,
659
- "requestCharacters": 332,
660
- "tools": [
661
- {
662
- "name": "search_many",
663
- "isError": false,
664
- "characters": 3796
665
- },
666
- {
667
- "name": "search_many",
668
- "isError": false,
669
- "characters": 2689
670
- }
671
- ]
672
- },
673
- {
674
- "index": 14,
675
- "status": "completed",
676
- "inputTokens": 7355,
677
- "outputTokens": 58,
678
- "cachedInputTokens": 1876,
679
- "requestCharacters": 32299,
680
- "tools": [
681
- {
682
- "name": "read_files",
683
- "isError": false,
684
- "characters": 6030
685
- }
686
- ]
687
- },
688
- {
689
- "index": 15,
690
- "status": "completed",
691
- "inputTokens": 9062,
692
- "outputTokens": 119,
693
- "cachedInputTokens": 1876,
694
- "requestCharacters": 39349,
695
- "tools": []
696
- }
697
- ],
698
- "inputTokens": 102877,
699
- "outputTokens": 1528,
700
- "cachedInputTokens": 47211,
701
- "modelCalls": 15,
702
- "compactions": [
703
- {
704
- "beforeStep": 4,
705
- "estimatedTokensBefore": 10798,
706
- "estimatedTokensAfter": 4805
707
- },
708
- {
709
- "beforeStep": 6,
710
- "estimatedTokensBefore": 13352,
711
- "estimatedTokensAfter": 5017
712
- },
713
- {
714
- "beforeStep": 8,
715
- "estimatedTokensBefore": 12607,
716
- "estimatedTokensAfter": 6489
717
- },
718
- {
719
- "beforeStep": 10,
720
- "estimatedTokensBefore": 12022,
721
- "estimatedTokensAfter": 4718
722
- },
723
- {
724
- "beforeStep": 13,
725
- "estimatedTokensBefore": 12616,
726
- "estimatedTokensAfter": 5123
727
- }
728
- ],
729
- "terminalDiagnostic": {
730
- "code": "EXECUTION_FAILED",
731
- "category": "execution",
732
- "retryable": false,
733
- "chain": [
734
- {
735
- "providerReason": null,
736
- "validationIssues": [],
737
- "status": null,
738
- "errorClass": "Object",
739
- "providerHints": [],
740
- "fingerprint": "62eab16f3e5ef3caebb5275363ae87ebc8e0f4b9bb23f9382c11f1fc18fef0e5",
741
- "reason": "budget"
742
- }
743
- ]
744
- },
745
- "thrownDiagnostic": {
746
- "code": "EXECUTION_FAILED",
747
- "category": "execution",
748
- "retryable": false,
749
- "chain": [
750
- {
751
- "providerReason": null,
752
- "validationIssues": [],
753
- "status": null,
754
- "errorClass": "HarnessExecutionError",
755
- "providerHints": [],
756
- "fingerprint": "62eab16f3e5ef3caebb5275363ae87ebc8e0f4b9bb23f9382c11f1fc18fef0e5",
757
- "reason": "budget"
758
- },
759
- {
760
- "providerReason": null,
761
- "validationIssues": [],
762
- "status": null,
763
- "errorClass": "GuardrailTriggeredError",
764
- "providerHints": [],
765
- "fingerprint": "62eab16f3e5ef3caebb5275363ae87ebc8e0f4b9bb23f9382c11f1fc18fef0e5",
766
- "reason": "budget"
767
- }
768
- ]
769
- },
770
- "guardrails": [],
771
- "toolCalls": 19,
772
- "patchSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
773
- "protectedFilesChanged": [],
774
- "gradingMethod": "empty-patch-no-solution",
775
- "officialResolved": false,
776
- "gradingDurationMs": 0.022833002731204033,
777
- "totalDurationMs": 37479.875040997285
778
- }
779
- ],
780
- "baseline": "external-matched-openai-2026-09-08.json",
781
- "treatment": {
782
- "maxTerminalVerificationRetries": 2,
783
- "changedSources": [
784
- "src/harness.ts",
785
- "scripts/swebench/zhivex-driver.ts"
786
- ]
787
- },
788
- "limitations": [
789
- "Two deliberately selected development cases, one repetition.",
790
- "Django stopped at input budget without invoking verifier; no live recovery was exercised.",
791
- "Recovery is established by deterministic tests, not this live score.",
792
- "No price rates supplied."
793
- ]
794
- }