@tangle-network/agent-eval 0.144.11 → 0.144.13

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 (378) hide show
  1. package/CHANGELOG.md +11 -0
  2. package/dist/agent-profile-B9_GGsG8.d.ts +84 -0
  3. package/dist/agent-profile-B9_GGsG8.d.ts.map +1 -0
  4. package/dist/{agent-profile-cell-BOP-iA9Q.d.ts → agent-profile-cell-BkcRDikH.d.ts} +2 -2
  5. package/dist/{agent-profile-cell-BOP-iA9Q.d.ts.map → agent-profile-cell-BkcRDikH.d.ts.map} +1 -1
  6. package/dist/analyst/index.d.ts +134 -16
  7. package/dist/analyst/index.d.ts.map +1 -1
  8. package/dist/analyst/index.js +364 -10
  9. package/dist/analyst/index.js.map +1 -1
  10. package/dist/backend-integrity-CsVin_Wb.d.ts +280 -0
  11. package/dist/backend-integrity-CsVin_Wb.d.ts.map +1 -0
  12. package/dist/{baseline-C-GocmIW.js → baseline-BhPRQBVn.js} +3 -2
  13. package/dist/{baseline-C-GocmIW.js.map → baseline-BhPRQBVn.js.map} +1 -1
  14. package/dist/{benchmark-CWeqGl7x.js → benchmark-BhT16ep9.js} +2 -2
  15. package/dist/{benchmark-CWeqGl7x.js.map → benchmark-BhT16ep9.js.map} +1 -1
  16. package/dist/{agent-profile-DPi7IZg7.d.ts → benchmark-Ceoan7vk.d.ts} +4 -91
  17. package/dist/benchmark-Ceoan7vk.d.ts.map +1 -0
  18. package/dist/{benchmark-command-BteMFN62.js → benchmark-command-CY6cbhOB.js} +36 -79
  19. package/dist/benchmark-command-CY6cbhOB.js.map +1 -0
  20. package/dist/benchmarks/index.d.ts +244 -2
  21. package/dist/benchmarks/index.d.ts.map +1 -0
  22. package/dist/benchmarks/index.js +733 -1
  23. package/dist/benchmarks/index.js.map +1 -0
  24. package/dist/builder-eval/index.d.ts +23 -2
  25. package/dist/builder-eval/index.d.ts.map +1 -1
  26. package/dist/builder-eval/index.js +227 -3
  27. package/dist/builder-eval/index.js.map +1 -1
  28. package/dist/campaign/index.d.ts +10 -8
  29. package/dist/campaign/index.js +9 -6
  30. package/dist/{campaign-C2TTzQII.js → campaign-BYjBAypg.js} +21 -634
  31. package/dist/campaign-BYjBAypg.js.map +1 -0
  32. package/dist/{canonical-D011XM8r.js → canonical-D-XsTQ6_.js} +2 -2
  33. package/dist/{canonical-D011XM8r.js.map → canonical-D-XsTQ6_.js.map} +1 -1
  34. package/dist/{index-CQTZ-4XN.d.ts → capture-fetch-DDvpjVRU.d.ts} +3 -3
  35. package/dist/capture-fetch-DDvpjVRU.d.ts.map +1 -0
  36. package/dist/{default-registry-BmktKy8r.js → chat-client-Bp4Ebuuc.js} +1276 -1276
  37. package/dist/chat-client-Bp4Ebuuc.js.map +1 -0
  38. package/dist/cli.js +2 -2
  39. package/dist/{client-C9gzZE59.d.ts → client-KUbGulm_.d.ts} +4 -4
  40. package/dist/{client-C9gzZE59.d.ts.map → client-KUbGulm_.d.ts.map} +1 -1
  41. package/dist/contract/index.d.ts +13 -390
  42. package/dist/contract/index.d.ts.map +1 -1
  43. package/dist/contract/index.js +18 -542
  44. package/dist/contract/index.js.map +1 -1
  45. package/dist/{cost-ledger-DMFxsLKr.js → cost-ledger-BSe92yAV.js} +3 -3
  46. package/dist/{cost-ledger-DMFxsLKr.js.map → cost-ledger-BSe92yAV.js.map} +1 -1
  47. package/dist/{cost-ledger-Bv_e8XHY.d.ts → cost-ledger-DbQdN3nO.d.ts} +2 -2
  48. package/dist/{cost-ledger-Bv_e8XHY.d.ts.map → cost-ledger-DbQdN3nO.d.ts.map} +1 -1
  49. package/dist/{counterfactual-CxmxAONP.d.ts → counterfactual--bpysZF0.d.ts} +2 -14
  50. package/dist/{counterfactual-CxmxAONP.d.ts.map → counterfactual--bpysZF0.d.ts.map} +1 -1
  51. package/dist/{counterfactual-CWPTrMH7.js → counterfactual-lDfCx0Uz.js} +3 -31
  52. package/dist/{counterfactual-CWPTrMH7.js.map → counterfactual-lDfCx0Uz.js.map} +1 -1
  53. package/dist/{dataset-C8xaLXdY.d.ts → dataset-CJjKqQfA.d.ts} +2 -8
  54. package/dist/dataset-CJjKqQfA.d.ts.map +1 -0
  55. package/dist/{default-registry-Bf8Woqmq.d.ts → default-registry-Di6HP6pG.d.ts} +7 -12
  56. package/dist/default-registry-Di6HP6pG.d.ts.map +1 -0
  57. package/dist/{analyze-runs-C30yljDJ.js → define-agent-eval-iqjT--ZZ.js} +542 -130
  58. package/dist/define-agent-eval-iqjT--ZZ.js.map +1 -0
  59. package/dist/define-agent-eval-sH24zBfM.d.ts +388 -0
  60. package/dist/define-agent-eval-sH24zBfM.d.ts.map +1 -0
  61. package/dist/descriptive-B5MwKfbf.js +144 -0
  62. package/dist/descriptive-B5MwKfbf.js.map +1 -0
  63. package/dist/{dspy-rlm-engine-DbTk4JdR.js → dspy-rlm-engine-B_qhSc21.js} +4 -4
  64. package/dist/{dspy-rlm-engine-DbTk4JdR.js.map → dspy-rlm-engine-B_qhSc21.js.map} +1 -1
  65. package/dist/effect-sizes-DiH8MGOH.js +82 -0
  66. package/dist/effect-sizes-DiH8MGOH.js.map +1 -0
  67. package/dist/{engine-3hL-XqwJ.d.ts → engine-otFpE2gF.d.ts} +10 -38
  68. package/dist/engine-otFpE2gF.d.ts.map +1 -0
  69. package/dist/{errors-CKPfb2aH.d.ts → errors-DEE6u6ot.d.ts} +2 -14
  70. package/dist/{errors-CKPfb2aH.d.ts.map → errors-DEE6u6ot.d.ts.map} +1 -1
  71. package/dist/{errors-D-LKuDhb.js → errors-Dngq5h35.js} +2 -8
  72. package/dist/{errors-D-LKuDhb.js.map → errors-Dngq5h35.js.map} +1 -1
  73. package/dist/{eval-campaign-B_7wcnav.js → eval-campaign-UB-usSQ2.js} +6 -6
  74. package/dist/{eval-campaign-B_7wcnav.js.map → eval-campaign-UB-usSQ2.js.map} +1 -1
  75. package/dist/{exact-types-DSFFpLLI.d.ts → exact-types-BH1twmAJ.d.ts} +2 -2
  76. package/dist/{exact-types-DSFFpLLI.d.ts.map → exact-types-BH1twmAJ.d.ts.map} +1 -1
  77. package/dist/experiment/index.d.ts +9 -6
  78. package/dist/experiment/index.d.ts.map +1 -1
  79. package/dist/experiment/index.js +11 -7
  80. package/dist/experiment/index.js.map +1 -1
  81. package/dist/experiment-tracker-C29gXM4B.js +269 -0
  82. package/dist/experiment-tracker-C29gXM4B.js.map +1 -0
  83. package/dist/{experiment-tracker-IMntXr6J.d.ts → experiment-tracker-DWHZBAYL.d.ts} +77 -121
  84. package/dist/experiment-tracker-DWHZBAYL.d.ts.map +1 -0
  85. package/dist/{external-optimizer-contracts-CZuJNcT5.d.ts → external-optimizer-contracts-lixrOZdX.d.ts} +3 -3
  86. package/dist/{external-optimizer-contracts-CZuJNcT5.d.ts.map → external-optimizer-contracts-lixrOZdX.d.ts.map} +1 -1
  87. package/dist/external-optimizer-process-BTiNB-RH.js +301 -0
  88. package/dist/external-optimizer-process-BTiNB-RH.js.map +1 -0
  89. package/dist/{single-run-lock-DFWHEB09.js → external-optimizer-subprocess-DrJ9hR8u.js} +144 -438
  90. package/dist/external-optimizer-subprocess-DrJ9hR8u.js.map +1 -0
  91. package/dist/extract-usage-BrQ8mCLX.js +155 -0
  92. package/dist/extract-usage-BrQ8mCLX.js.map +1 -0
  93. package/dist/{failure-cluster-CqcvCcdR.d.ts → failure-cluster-BLURuWG4.d.ts} +2 -3
  94. package/dist/failure-cluster-BLURuWG4.d.ts.map +1 -0
  95. package/dist/{feedback-trajectory-CacHpxsp.d.ts → feedback-trajectory-juOozjAc.d.ts} +5 -36
  96. package/dist/feedback-trajectory-juOozjAc.d.ts.map +1 -0
  97. package/dist/fuzz.d.ts +2 -2
  98. package/dist/fuzz.js +2 -2
  99. package/dist/hosted/index.d.ts +3 -3
  100. package/dist/{index-CvSN3IG1.d.ts → index-B8Ui1mr1.d.ts} +2 -2
  101. package/dist/{index-CvSN3IG1.d.ts.map → index-B8Ui1mr1.d.ts.map} +1 -1
  102. package/dist/{skill-usage-CO9OLRBx.d.ts → index-BWDrSVfw.d.ts} +11 -136
  103. package/dist/index-BWDrSVfw.d.ts.map +1 -0
  104. package/dist/index-Ba3YrbAL.d.ts +1 -0
  105. package/dist/{index-BnEuDAK2.d.ts → index-COQYtuRF.d.ts} +3 -3
  106. package/dist/{index-BnEuDAK2.d.ts.map → index-COQYtuRF.d.ts.map} +1 -1
  107. package/dist/{index-C3ssXVLv.d.ts → index-CvjYbU0D.d.ts} +2 -2
  108. package/dist/{index-C3ssXVLv.d.ts.map → index-CvjYbU0D.d.ts.map} +1 -1
  109. package/dist/{index-DPPGNJ_R.d.ts → index-DSmEylT9.d.ts} +17 -115
  110. package/dist/index-DSmEylT9.d.ts.map +1 -0
  111. package/dist/index.d.ts +2397 -5308
  112. package/dist/index.d.ts.map +1 -1
  113. package/dist/index.js +5914 -10496
  114. package/dist/index.js.map +1 -1
  115. package/dist/{insight-report-CgX_s0Ez.d.ts → insight-report-CRi-Ufrj.d.ts} +4 -4
  116. package/dist/{insight-report-CgX_s0Ez.d.ts.map → insight-report-CRi-Ufrj.d.ts.map} +1 -1
  117. package/dist/{integrity-MLzHOfV9.js → integrity-Cy9WHAtb.js} +2 -2
  118. package/dist/{integrity-MLzHOfV9.js.map → integrity-Cy9WHAtb.js.map} +1 -1
  119. package/dist/{integrity-DY6tIbl0.js → integrity-DysDBWDu.js} +2 -2
  120. package/dist/{integrity-DY6tIbl0.js.map → integrity-DysDBWDu.js.map} +1 -1
  121. package/dist/{integrity-BuqEKu-x.d.ts → integrity-OrcI9Nau.d.ts} +3 -3
  122. package/dist/{integrity-BuqEKu-x.d.ts.map → integrity-OrcI9Nau.d.ts.map} +1 -1
  123. package/dist/internal-BDHPCnjk.js +230 -0
  124. package/dist/internal-BDHPCnjk.js.map +1 -0
  125. package/dist/judge-calibration-C5CbMYce.d.ts +117 -0
  126. package/dist/judge-calibration-C5CbMYce.d.ts.map +1 -0
  127. package/dist/judge-calibration-DZkWrm5H.js +317 -0
  128. package/dist/judge-calibration-DZkWrm5H.js.map +1 -0
  129. package/dist/{kind-factory-B8-r8-y8.js → kind-factory-CPmSd58s.js} +208 -208
  130. package/dist/{kind-factory-B8-r8-y8.js.map → kind-factory-CPmSd58s.js.map} +1 -1
  131. package/dist/ledger-core/index.d.ts +1 -1
  132. package/dist/ledger-core/index.js +2 -2
  133. package/dist/{ledger-core-DXZIqu17.js → ledger-core-BmZt19oQ.js} +110 -110
  134. package/dist/{ledger-core-DXZIqu17.js.map → ledger-core-BmZt19oQ.js.map} +1 -1
  135. package/dist/{llm-client-DzvMUsS_.js → llm-client-d0-2TT1g.js} +4 -64
  136. package/dist/{llm-client-DzvMUsS_.js.map → llm-client-d0-2TT1g.js.map} +1 -1
  137. package/dist/{skillopt-optimization-method-CQdVeM8k.js → llm-judge-dZ8P6nGI.js} +3349 -5401
  138. package/dist/llm-judge-dZ8P6nGI.js.map +1 -0
  139. package/dist/matrix/index.d.ts +2 -2
  140. package/dist/meta-eval/index.d.ts +3 -3
  141. package/dist/meta-eval/index.js +3 -3
  142. package/dist/{metrics-C9YY1OcL.js → metrics-Cl0L1KUy.js} +2 -108
  143. package/dist/{metrics-C9YY1OcL.js.map → metrics-Cl0L1KUy.js.map} +1 -1
  144. package/dist/{mint-B2O60ACG.js → mint-Dj9Ww_3I.js} +4 -4
  145. package/dist/{mint-B2O60ACG.js.map → mint-Dj9Ww_3I.js.map} +1 -1
  146. package/dist/{multi-layer-verifier-DnAqwl0h.d.ts → multi-layer-verifier-DIguZc8Z.d.ts} +3 -3
  147. package/dist/{multi-layer-verifier-DnAqwl0h.d.ts.map → multi-layer-verifier-DIguZc8Z.d.ts.map} +1 -1
  148. package/dist/multiplicity-DIWHvysC.d.ts +43 -0
  149. package/dist/multiplicity-DIWHvysC.d.ts.map +1 -0
  150. package/dist/multishot/index.d.ts +3 -3
  151. package/dist/multishot/index.js +1 -1
  152. package/dist/openapi.json +1 -1
  153. package/dist/{opencode-sqlite-8r6WUfHc.js → opencode-sqlite-DJWAXLms.js} +2 -2
  154. package/dist/{opencode-sqlite-8r6WUfHc.js.map → opencode-sqlite-DJWAXLms.js.map} +1 -1
  155. package/dist/package-version-D7lQHt_-.js +34 -0
  156. package/dist/package-version-D7lQHt_-.js.map +1 -0
  157. package/dist/paired-arms-D-XRF_fy.js +1045 -0
  158. package/dist/paired-arms-D-XRF_fy.js.map +1 -0
  159. package/dist/paired-promotion-decision-CGzg0cI_.d.ts +251 -0
  160. package/dist/paired-promotion-decision-CGzg0cI_.d.ts.map +1 -0
  161. package/dist/paired-tests-BHIhYVdu.js +213 -0
  162. package/dist/paired-tests-BHIhYVdu.js.map +1 -0
  163. package/dist/pareto-BqNW3LJR.d.ts +117 -0
  164. package/dist/pareto-BqNW3LJR.d.ts.map +1 -0
  165. package/dist/pipelines/index.d.ts +3 -64
  166. package/dist/pipelines/index.d.ts.map +1 -1
  167. package/dist/pipelines/index.js +4 -284
  168. package/dist/pipelines/index.js.map +1 -1
  169. package/dist/power-and-mde-CHIrXJll.js +195 -0
  170. package/dist/power-and-mde-CHIrXJll.js.map +1 -0
  171. package/dist/{promotion-policy-CrLrmys8.js → power-preflight-DEw-uC7q.js} +4 -184
  172. package/dist/power-preflight-DEw-uC7q.js.map +1 -0
  173. package/dist/pre-registration-CZwSFQS4.d.ts +577 -0
  174. package/dist/pre-registration-CZwSFQS4.d.ts.map +1 -0
  175. package/dist/{prime-protocol-BfSalTfR.js → prime-protocol-6tZTVsWm.js} +72 -21
  176. package/dist/prime-protocol-6tZTVsWm.js.map +1 -0
  177. package/dist/produced-state-DU79a81m.js +586 -0
  178. package/dist/produced-state-DU79a81m.js.map +1 -0
  179. package/dist/profile-cell.d.ts +1 -1
  180. package/dist/profile-cell.js +1 -1
  181. package/dist/promotion-policy-D0nPhkSy.d.ts +134 -0
  182. package/dist/promotion-policy-D0nPhkSy.d.ts.map +1 -0
  183. package/dist/promotion-policy-xzA40Evo.js +186 -0
  184. package/dist/promotion-policy-xzA40Evo.js.map +1 -0
  185. package/dist/{skillopt-optimization-method-USDKhxSA.d.ts → provenance-CCdxgLDT.d.ts} +57 -478
  186. package/dist/provenance-CCdxgLDT.d.ts.map +1 -0
  187. package/dist/registry-oJeeI4-a.d.ts +178 -0
  188. package/dist/registry-oJeeI4-a.d.ts.map +1 -0
  189. package/dist/{release-report-DA2BCu5a.d.ts → release-confidence-BFRE5WSp.d.ts} +180 -112
  190. package/dist/release-confidence-BFRE5WSp.d.ts.map +1 -0
  191. package/dist/{release-report-BUYmoKo2.js → release-confidence-CxDuiAev.js} +125 -217
  192. package/dist/release-confidence-CxDuiAev.js.map +1 -0
  193. package/dist/reporting.d.ts +6 -5
  194. package/dist/reporting.js +7 -5
  195. package/dist/{researcher-BchpD55R.d.ts → researcher-DaN4GST-.d.ts} +7 -40
  196. package/dist/{researcher-BchpD55R.d.ts.map → researcher-DaN4GST-.d.ts.map} +1 -1
  197. package/dist/{reward-hacking-BDToousL.js → reward-hacking-DNgjilrV.js} +3 -3
  198. package/dist/reward-hacking-DNgjilrV.js.map +1 -0
  199. package/dist/{reward-hacking-BI0OMAlo.d.ts → reward-hacking-DSSTuI9r.d.ts} +5 -5
  200. package/dist/{reward-hacking-BI0OMAlo.d.ts.map → reward-hacking-DSSTuI9r.d.ts.map} +1 -1
  201. package/dist/rl.d.ts +7 -7
  202. package/dist/rl.js +11 -10
  203. package/dist/rl.js.map +1 -1
  204. package/dist/rollout/index.d.ts +2 -2
  205. package/dist/rollout/index.js +4 -4
  206. package/dist/{rollout-DRrksrcV.js → rollout-BWtw0I_6.js} +3 -3
  207. package/dist/{rollout-DRrksrcV.js.map → rollout-BWtw0I_6.js.map} +1 -1
  208. package/dist/{rubric-predictive-validity-DvkjPYCe.d.ts → rubric-predictive-validity-BgxtKe4G.d.ts} +2 -2
  209. package/dist/{rubric-predictive-validity-DvkjPYCe.d.ts.map → rubric-predictive-validity-BgxtKe4G.d.ts.map} +1 -1
  210. package/dist/{rubric-predictive-validity-BRR632r1.js → rubric-predictive-validity-Cwwyd7ah.js} +2 -2
  211. package/dist/{rubric-predictive-validity-BRR632r1.js.map → rubric-predictive-validity-Cwwyd7ah.js.map} +1 -1
  212. package/dist/{run-record-BmSPWXJR.js → run-record-BvHPVS-i.js} +3 -3
  213. package/dist/{run-record-BmSPWXJR.js.map → run-record-BvHPVS-i.js.map} +1 -1
  214. package/dist/{run-record-CF4Dwpxr.d.ts → run-record-CKiihE6f.d.ts} +4 -4
  215. package/dist/{run-record-CF4Dwpxr.d.ts.map → run-record-CKiihE6f.d.ts.map} +1 -1
  216. package/dist/{proposal-findings-2GIUo1et.js → run-score-lDzV0X8j.js} +31 -31
  217. package/dist/run-score-lDzV0X8j.js.map +1 -0
  218. package/dist/sandbox-harness-BlSOu4LX.d.ts +70 -0
  219. package/dist/sandbox-harness-BlSOu4LX.d.ts.map +1 -0
  220. package/dist/{schema-Cef2cFmb2.d.ts → schema-Cef2cFmb.d.ts} +1 -1
  221. package/dist/schema-Cef2cFmb.d.ts.map +1 -0
  222. package/dist/semantic-concept-judge-BiJxScqe.js +406 -0
  223. package/dist/semantic-concept-judge-BiJxScqe.js.map +1 -0
  224. package/dist/{sequential-D-BLJBKU.js → sequential-C458DXNf.js} +4 -3
  225. package/dist/{sequential-D-BLJBKU.js.map → sequential-C458DXNf.js.map} +1 -1
  226. package/dist/sequential-eprocess-CbUt2htw.js +83 -0
  227. package/dist/sequential-eprocess-CbUt2htw.js.map +1 -0
  228. package/dist/series-convergence-D1cL1f-4.d.ts +129 -0
  229. package/dist/series-convergence-D1cL1f-4.d.ts.map +1 -0
  230. package/dist/{server-iu0ede49.js → server-ulsOdrTI.js} +5 -21
  231. package/dist/server-ulsOdrTI.js.map +1 -0
  232. package/dist/skillopt-optimization-method-C5cotF4E.d.ts +427 -0
  233. package/dist/skillopt-optimization-method-C5cotF4E.d.ts.map +1 -0
  234. package/dist/skillopt-optimization-method-jjdnc3YK.js +1999 -0
  235. package/dist/skillopt-optimization-method-jjdnc3YK.js.map +1 -0
  236. package/dist/{statistical-heldout-TQ-4CYiN.d.ts → statistical-heldout-DoFF5KJX.d.ts} +5 -278
  237. package/dist/statistical-heldout-DoFF5KJX.d.ts.map +1 -0
  238. package/dist/{store-otlp-Dw8PPIlL.js → store-otlp-CsptLYpN.js} +3 -3
  239. package/dist/{store-otlp-Dw8PPIlL.js.map → store-otlp-CsptLYpN.js.map} +1 -1
  240. package/dist/store-tool-spans-Cq9mFd-q.js +667 -0
  241. package/dist/store-tool-spans-Cq9mFd-q.js.map +1 -0
  242. package/dist/store-tool-spans-DbDLOBOb.d.ts +342 -0
  243. package/dist/store-tool-spans-DbDLOBOb.d.ts.map +1 -0
  244. package/dist/student-t-CvBq2mve.js +38 -0
  245. package/dist/student-t-CvBq2mve.js.map +1 -0
  246. package/dist/{summary-report-Lf-5I7xh.js → summary-report-Blysd6Z2.js} +6 -3
  247. package/dist/{summary-report-Lf-5I7xh.js.map → summary-report-Blysd6Z2.js.map} +1 -1
  248. package/dist/{summary-report-BNR7DWTj.d.ts → summary-report-CaL-Hnxt.d.ts} +5 -5
  249. package/dist/{summary-report-BNR7DWTj.d.ts.map → summary-report-CaL-Hnxt.d.ts.map} +1 -1
  250. package/dist/supervisor-run/index.d.ts +391 -3
  251. package/dist/supervisor-run/index.d.ts.map +1 -0
  252. package/dist/supervisor-run/index.js +1689 -2
  253. package/dist/{supervisor-run-D_sokXcO.js.map → supervisor-run/index.js.map} +1 -1
  254. package/dist/{extract-usage-CdZdoj1s.js → task-failure-attributes-CpQ4y5RD.js} +5 -157
  255. package/dist/task-failure-attributes-CpQ4y5RD.js.map +1 -0
  256. package/dist/{tool-groups-CZPGGlHf.d.ts → tool-groups-Cteb03Ps.d.ts} +3 -3
  257. package/dist/tool-groups-Cteb03Ps.d.ts.map +1 -0
  258. package/dist/tool-waste-BDdBZG1F.js +803 -0
  259. package/dist/tool-waste-BDdBZG1F.js.map +1 -0
  260. package/dist/tool-waste-DjRDEsuI.d.ts +128 -0
  261. package/dist/tool-waste-DjRDEsuI.d.ts.map +1 -0
  262. package/dist/trace-repair/index.d.ts +14 -5
  263. package/dist/trace-repair/index.d.ts.map +1 -1
  264. package/dist/trace-repair/index.js +35 -7
  265. package/dist/trace-repair/index.js.map +1 -1
  266. package/dist/traces.d.ts +406 -7
  267. package/dist/traces.d.ts.map +1 -0
  268. package/dist/traces.js +1011 -10
  269. package/dist/traces.js.map +1 -0
  270. package/dist/trajectory-replay/index.d.ts +16 -3
  271. package/dist/trajectory-replay/index.d.ts.map +1 -1
  272. package/dist/trajectory-replay/index.js +52 -5
  273. package/dist/trajectory-replay/index.js.map +1 -1
  274. package/dist/types-BEPZc6eo.d.ts +93 -0
  275. package/dist/types-BEPZc6eo.d.ts.map +1 -0
  276. package/dist/{usage-receipt-t7vAzCRQ.js → types-BI4fT3HN.js} +67 -67
  277. package/dist/types-BI4fT3HN.js.map +1 -0
  278. package/dist/{types-BnjdJ70P.d.ts → types-BZ59Ahr8.d.ts} +5 -5
  279. package/dist/{types-BnjdJ70P.d.ts.map → types-BZ59Ahr8.d.ts.map} +1 -1
  280. package/dist/{types-D216SgwM.d.ts → types-Cx3YUh2r.d.ts} +4 -241
  281. package/dist/types-Cx3YUh2r.d.ts.map +1 -0
  282. package/dist/{types-BvDKaULh.d.ts → types-DLQx4mKU.d.ts} +4 -4
  283. package/dist/{types-BvDKaULh.d.ts.map → types-DLQx4mKU.d.ts.map} +1 -1
  284. package/dist/{types-D4mog56g.d.ts → types-yLK8gXE9.d.ts} +2 -2
  285. package/dist/{types-D4mog56g.d.ts.map → types-yLK8gXE9.d.ts.map} +1 -1
  286. package/dist/verdict-BndeTAh_.js +61 -0
  287. package/dist/verdict-BndeTAh_.js.map +1 -0
  288. package/dist/verdict-E4eRNf7-.d.ts +392 -0
  289. package/dist/verdict-E4eRNf7-.d.ts.map +1 -0
  290. package/dist/{verdict-cache-BCcOh0kF.js → verdict-cache-mZf5FEiY.js} +3 -55
  291. package/dist/{verdict-cache-BCcOh0kF.js.map → verdict-cache-mZf5FEiY.js.map} +1 -1
  292. package/dist/wire/index.d.ts +3 -3
  293. package/dist/wire/index.d.ts.map +1 -1
  294. package/dist/wire/index.js +1 -1
  295. package/docs/charter.md +3 -3
  296. package/docs/control-runtime.md +3 -42
  297. package/docs/experiment.md +0 -1
  298. package/docs/feature-guide.md +2 -2
  299. package/docs/trace-repair-grader.md +1 -0
  300. package/docs/trajectory-replay.md +1 -0
  301. package/docs/verdicts.md +43 -0
  302. package/docs/verification-strategies.md +3 -2
  303. package/package.json +6 -11
  304. package/dist/agent-profile-DPi7IZg7.d.ts.map +0 -1
  305. package/dist/analyze-runs-C30yljDJ.js.map +0 -1
  306. package/dist/baseline-CavEbRyH.d.ts +0 -136
  307. package/dist/baseline-CavEbRyH.d.ts.map +0 -1
  308. package/dist/benchmark-command-BteMFN62.js.map +0 -1
  309. package/dist/benchmarks-Dzs8CKb1.js +0 -755
  310. package/dist/benchmarks-Dzs8CKb1.js.map +0 -1
  311. package/dist/campaign-C2TTzQII.js.map +0 -1
  312. package/dist/completion-verifier-DJA5BhPb.d.ts +0 -414
  313. package/dist/completion-verifier-DJA5BhPb.d.ts.map +0 -1
  314. package/dist/control.d.ts +0 -3
  315. package/dist/control.js +0 -2
  316. package/dist/dataset-C8xaLXdY.d.ts.map +0 -1
  317. package/dist/default-registry-Bf8Woqmq.d.ts.map +0 -1
  318. package/dist/default-registry-BmktKy8r.js.map +0 -1
  319. package/dist/engine-3hL-XqwJ.d.ts.map +0 -1
  320. package/dist/experiment-tracker-CnRICnMl.js +0 -500
  321. package/dist/experiment-tracker-CnRICnMl.js.map +0 -1
  322. package/dist/experiment-tracker-IMntXr6J.d.ts.map +0 -1
  323. package/dist/extract-usage-CdZdoj1s.js.map +0 -1
  324. package/dist/failure-cluster-CqcvCcdR.d.ts.map +0 -1
  325. package/dist/feedback-trajectory-CacHpxsp.d.ts.map +0 -1
  326. package/dist/index-BZ3-y4YL.d.ts +0 -391
  327. package/dist/index-BZ3-y4YL.d.ts.map +0 -1
  328. package/dist/index-CQTZ-4XN.d.ts.map +0 -1
  329. package/dist/index-DPPGNJ_R.d.ts.map +0 -1
  330. package/dist/index-YE4KdKbO2.d.ts +0 -335
  331. package/dist/index-YE4KdKbO2.d.ts.map +0 -1
  332. package/dist/paired-arms-iZ08VFMN.js +0 -260
  333. package/dist/paired-arms-iZ08VFMN.js.map +0 -1
  334. package/dist/paired-promotion-decision-B6zJ3gYM.d.ts +0 -114
  335. package/dist/paired-promotion-decision-B6zJ3gYM.d.ts.map +0 -1
  336. package/dist/prime-protocol-BfSalTfR.js.map +0 -1
  337. package/dist/promotion-policy-Ckjhzg_4.d.ts +0 -289
  338. package/dist/promotion-policy-Ckjhzg_4.d.ts.map +0 -1
  339. package/dist/promotion-policy-CrLrmys8.js.map +0 -1
  340. package/dist/proposal-findings-2GIUo1et.js.map +0 -1
  341. package/dist/propose-review-control-dSNPjFUH.js +0 -1458
  342. package/dist/propose-review-control-dSNPjFUH.js.map +0 -1
  343. package/dist/release-report-BUYmoKo2.js.map +0 -1
  344. package/dist/release-report-DA2BCu5a.d.ts.map +0 -1
  345. package/dist/replay-CohS93nE.js +0 -1859
  346. package/dist/replay-CohS93nE.js.map +0 -1
  347. package/dist/replay-DbhZ4Ked.d.ts +0 -834
  348. package/dist/replay-DbhZ4Ked.d.ts.map +0 -1
  349. package/dist/reward-hacking-BDToousL.js.map +0 -1
  350. package/dist/run-evidence-Dhi3C81V.d.ts +0 -225
  351. package/dist/run-evidence-Dhi3C81V.d.ts.map +0 -1
  352. package/dist/schema-Cef2cFmb2.d.ts.map +0 -1
  353. package/dist/semantic-concept-judge-D1z-KepS.js +0 -767
  354. package/dist/semantic-concept-judge-D1z-KepS.js.map +0 -1
  355. package/dist/series-convergence-ofsqPWhs.d.ts +0 -35
  356. package/dist/series-convergence-ofsqPWhs.d.ts.map +0 -1
  357. package/dist/server-iu0ede49.js.map +0 -1
  358. package/dist/single-run-lock-DFWHEB09.js.map +0 -1
  359. package/dist/skill-usage-CO9OLRBx.d.ts.map +0 -1
  360. package/dist/skillopt-optimization-method-CQdVeM8k.js.map +0 -1
  361. package/dist/skillopt-optimization-method-USDKhxSA.d.ts.map +0 -1
  362. package/dist/statistical-heldout-TQ-4CYiN.d.ts.map +0 -1
  363. package/dist/statistics-ByxzSiOM.js +0 -2212
  364. package/dist/statistics-ByxzSiOM.js.map +0 -1
  365. package/dist/statistics-D6Uebe_4.d.ts +0 -968
  366. package/dist/statistics-D6Uebe_4.d.ts.map +0 -1
  367. package/dist/supervisor-run-D_sokXcO.js +0 -1690
  368. package/dist/test-graded-scenario-D1TaI2va.d.ts +0 -141
  369. package/dist/test-graded-scenario-D1TaI2va.d.ts.map +0 -1
  370. package/dist/test-graded-scenario-JHcKQNpq.js +0 -318
  371. package/dist/test-graded-scenario-JHcKQNpq.js.map +0 -1
  372. package/dist/tool-groups-CZPGGlHf.d.ts.map +0 -1
  373. package/dist/tool-use-metrics-DEGMKycK.js +0 -370
  374. package/dist/tool-use-metrics-DEGMKycK.js.map +0 -1
  375. package/dist/types-D216SgwM.d.ts.map +0 -1
  376. package/dist/usage-receipt-t7vAzCRQ.js.map +0 -1
  377. package/dist/verdict-DExhxfgR.d.ts +0 -201
  378. package/dist/verdict-DExhxfgR.d.ts.map +0 -1
@@ -1,201 +0,0 @@
1
- //#region src/verification-strategy.d.ts
2
- /**
3
- * Verification-strategy family — the shared vocabulary for "what certified
4
- * this result", covering tasks with an answer key and tasks without one.
5
- *
6
- * Charter Wave 5 (docs/charter.md): an unsolved problem has no held-out
7
- * suite by definition, so the held-out suite must be ONE member of a
8
- * strategy family, not the family itself. This module carries the taxonomy
9
- * and the port shape only. A checker is an injected executable boundary
10
- * that returns a typed outcome, so a consumer binds its own kernel (a Lean
11
- * toolchain, a metamorphic harness, a replication runner) and this package
12
- * never chooses one. Instruments, never methods.
13
- *
14
- * Full family doc, per-member failure modes in prose, and the BCWW pilot
15
- * worked example: docs/verification-strategies.md.
16
- *
17
- * Leaf module: no imports from this package. `src/verdict.ts` and
18
- * `src/rl/verifiable-reward.ts` both build on these types.
19
- */
20
- /**
21
- * One member of the verification-strategy family.
22
- *
23
- * Every member has a documented failure mode — the specific way it can
24
- * certify a wrong result. A consumer that reads a certification must weigh
25
- * the member's failure mode, not treat "certified" as one bit.
26
- *
27
- * - `'compile'` — typecheck / build / lint passed. Deterministic.
28
- * Failure mode: code that compiles is not code that is correct; the
29
- * weakest deterministic member.
30
- * - `'test'` — unit / integration / held-out suite pass-rate. Deterministic.
31
- * Failure mode: assumes an answer key exists; certifies nothing outside
32
- * the suite's coverage, and a stubbed integration reports green.
33
- * - `'schema'` — structured output validates. Deterministic.
34
- * Failure mode: shape is not meaning; a well-formed wrong answer passes.
35
- * - `'sandbox'` — sandbox execution exit code. Deterministic.
36
- * Failure mode: an exit code compresses the whole run to one bit; a
37
- * faked success exits 0.
38
- * - `'judge'` — LLM judge score. Probabilistic.
39
- * Failure mode: drifts across model versions and is Goodhart-gameable by
40
- * the graded policy.
41
- * - `'composite'` — weighted blend across members. Determinism inherited
42
- * from its members.
43
- * Failure mode: scalar collapse — the blend hides which member carried
44
- * the score and which failed.
45
- * - `'proof-kernel'` — a proof assistant's kernel accepted a formal proof.
46
- * Deterministic.
47
- * Failure mode: the formalization gap. The kernel certifies the formal
48
- * statement, never that the formal statement matches the informal claim;
49
- * a proved theorem about the wrong statement is a wrong result with a
50
- * valid certificate. Discharge it with the blind statement-equivalence
51
- * protocol (`src/equivalence-check.ts`).
52
- * - `'invariant'` — invariant / metamorphic properties held on the result.
53
- * Deterministic.
54
- * Failure mode: weak invariants pass everything. An invariant set earns
55
- * weight only through seeded-bug calibration — a mutation the set
56
- * provably rejects; an uncalibrated set is a rubber stamp.
57
- * - `'replication'` — an independent re-execution reproduced the result
58
- * from pinned inputs. Deterministic given the pins.
59
- * Failure mode: replication re-runs the method, so it catches
60
- * nondeterminism and environment drift, never a methodological error the
61
- * method itself carries.
62
- * - `'agreement'` — independently-derived results agree. Probabilistic.
63
- * Failure mode: the shared blind spot. Derivers that share training
64
- * corpora, priors, or the same misread of the source agree for the same
65
- * wrong reason; the blindness provenance of each arm is what the
66
- * certificate rests on.
67
- */
68
- type VerificationStrategySource = 'compile' | 'test' | 'schema' | 'sandbox' | 'judge' | 'composite' | 'proof-kernel' | 'invariant' | 'replication' | 'agreement';
69
- /**
70
- * What a strategy member can honestly claim about its outcomes.
71
- *
72
- * `determinism: 'inherited'` exists only for `'composite'`: a blend takes
73
- * the class of its members. The reward axis stays binary
74
- * (`VerifiableReward.determinism`); this registry describes the family,
75
- * it never scores a run.
76
- */
77
- interface VerificationStrategyProfile {
78
- /** Determinism class a well-formed checker of this member can claim. */
79
- determinism: 'deterministic' | 'probabilistic' | 'inherited';
80
- /** The documented way this member certifies a wrong result. */
81
- failureMode: string;
82
- }
83
- /**
84
- * The family registry. `Record` over the union keeps it exhaustive: adding
85
- * a member to `VerificationStrategySource` without a profile here fails to
86
- * compile. The failure mode travels with the taxonomy so a reader of a
87
- * certification can surface it without this package's docs at hand.
88
- */
89
- declare const VERIFICATION_STRATEGIES: Record<VerificationStrategySource, VerificationStrategyProfile>;
90
- /** Every family member, derived from the registry so it cannot drift. */
91
- declare const VERIFICATION_STRATEGY_SOURCES: readonly VerificationStrategySource[];
92
- /**
93
- * Exact identity of the thing that checked. "lean4 4.33.0 with Mathlib
94
- * db584cd6d46c" and "lean4, some version" are different certificates: the
95
- * first is re-runnable, the second is a story.
96
- */
97
- interface CheckerIdentity {
98
- /** Toolchain or model name, e.g. `'lean4'` or a judge model id. */
99
- name: string;
100
- /** Exact version or snapshot, e.g. `'4.33.0'` or a model snapshot id. */
101
- version: string;
102
- /**
103
- * Content pins the version alone does not capture, e.g.
104
- * `{ mathlib: 'db584cd6d46c' }`. Keyed by dependency name.
105
- */
106
- pins?: Record<string, string>;
107
- }
108
- /**
109
- * Typed outcome of an external checker call. Callers MUST inspect
110
- * `succeeded` before touching `value`; a failed check keeps its full error
111
- * text instead of collapsing to null.
112
- */
113
- type CheckerOutcome<Result> = {
114
- succeeded: true;
115
- value: Result;
116
- } | {
117
- succeeded: false;
118
- error: string;
119
- };
120
- /**
121
- * The port shape every strategy binds through: an injected executable
122
- * boundary. This package defines the port and refuses to ship any
123
- * implementation behind it — the consumer binds its own kernel, invariant
124
- * harness, replication runner, or judge, and the binding carries its own
125
- * identity and determinism claim.
126
- *
127
- * A checker that ran but could not decide returns
128
- * `{ succeeded: false, error }` with the reason; `succeeded: true` is
129
- * reserved for a discharged check.
130
- */
131
- interface StrategyChecker<Input, Result> {
132
- /** Which family member this checker discharges obligations for. */
133
- strategy: VerificationStrategySource;
134
- identity: CheckerIdentity;
135
- /** Determinism class this checker claims for its outcomes. */
136
- determinism: 'deterministic' | 'probabilistic';
137
- check(input: Input): Promise<CheckerOutcome<Result>>;
138
- }
139
- //#endregion
140
- //#region src/verdict.d.ts
141
- /**
142
- * What certified a verdict — the epistemics a bare `valid` + `score` pair
143
- * cannot carry. A kernel-checked proof and an LLM judge can produce the
144
- * same `{ valid: true, score: 1 }`; this record is what tells them apart.
145
- *
146
- * Every field is required on purpose. A certification without a checker
147
- * identity or an evidence digest is exactly the unverifiable claim this
148
- * record exists to kill.
149
- */
150
- interface VerdictCertification {
151
- /**
152
- * Which verification-strategy member produced the certificate. Each
153
- * member's failure mode is documented on the family type and in
154
- * `VERIFICATION_STRATEGIES` (src/verification-strategy.ts).
155
- */
156
- strategy: VerificationStrategySource;
157
- /**
158
- * Exact checker identity, e.g. `{ name: 'lean4', version: '4.33.0',
159
- * pins: { mathlib: 'db584cd6d46c' } }` for a kernel, or a judge model
160
- * id + snapshot for a judge.
161
- */
162
- checker: CheckerIdentity;
163
- /**
164
- * Every step the certificate rests on that the checker did NOT verify,
165
- * e.g. 'diagonal embedding argued in prose' or 'agreement arms share
166
- * training-corpus blind spots'. An empty array is the producer's
167
- * explicit, auditable claim that there are none — it is not a default.
168
- */
169
- assumptions: string[];
170
- /** Digest (e.g. sha256) of the evidence artifact the checker consumed or produced. */
171
- evidenceDigest: string;
172
- }
173
- /**
174
- * Minimal verdict shape — `valid` + `score` are required; `scores` +
175
- * `notes` are optional surface. Validators that need richer shapes
176
- * parameterise `Validator<Output, MyVerdict>` with their own type.
177
- *
178
- * Need structured extras? Extend DefaultVerdict with typed fields — never
179
- * serialize extras into `notes`.
180
- */
181
- interface DefaultVerdict {
182
- /** Whether the output meets the validator's pass criteria. */
183
- valid: boolean;
184
- /** Aggregate score in [0, 1]. Drivers use this for winner selection. */
185
- score: number;
186
- /** Per-dimension scores. Free-form; weighted into `score` by the validator. */
187
- scores?: Record<string, number>;
188
- /** Human-readable rationale; surfaces in trace + final-result `winner.verdict`. */
189
- notes?: string;
190
- /**
191
- * What certified this verdict, when anything did. Absent = uncertified:
192
- * scored, but no strategy vouches for it. A consumer that cares reads
193
- * `certification.strategy` to tell a kernel-checked verdict from a
194
- * judge-scored one; a consumer that does not read it sees the exact
195
- * verdict it always saw.
196
- */
197
- certification?: VerdictCertification;
198
- }
199
- //#endregion
200
- export { StrategyChecker as a, VerificationStrategyProfile as c, CheckerOutcome as i, VerificationStrategySource as l, VerdictCertification as n, VERIFICATION_STRATEGIES as o, CheckerIdentity as r, VERIFICATION_STRATEGY_SOURCES as s, DefaultVerdict as t };
201
- //# sourceMappingURL=verdict-DExhxfgR.d.ts.map
@@ -1 +0,0 @@
1
- {"version":3,"file":"verdict-DExhxfgR.d.ts","names":[],"sources":["../src/verification-strategy.ts","../src/verdict.ts"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;KAmEY;;;;;;;;;UAoBK;;EAEf;;EAEA;;;;;;;;cASW,yBAAyB,OACpC,4BACA;;cAkDW,wCAEC;;;;;;UAOG;;EAEf;;EAEA;;;;;EAKA,OAAO;;;;;;;KAQG,eAAe;EACrB;EAAiB,OAAO;;EACxB;EAAkB;;;;;;;;;;;;;UAaP,gBAAgB,OAAO;;EAEtC,UAAU;EACV,UAAU;;EAEV;EACA,MAAM,OAAO,QAAQ,QAAQ,eAAe;;;;;;;;;;;;;UC5K7B;;;;;;EAMf,UAAU;;;;;;EAMV,SAAS;;;;;;;EAOT;;EAEA;;;;;;;;;;UAWe;;EAEf;;EAEA;;EAEA,SAAS;;EAET;;;;;;;;EAQA,gBAAgB"}