@tangle-network/agent-eval 0.216.1 → 0.217.1

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 (521) hide show
  1. package/CHANGELOG.md +102 -57
  2. package/README.md +2 -2
  3. package/dist/adapters/http.d.ts +2 -2
  4. package/dist/{agent-profile-VFcQncbW.d.ts → agent-profile-DTt8uGP9.d.ts} +4 -17
  5. package/dist/agent-profile-DTt8uGP9.d.ts.map +1 -0
  6. package/dist/{agent-profile-cell-BK8mlqlq.js → agent-profile-cell-B1V1gUZs.js} +4 -53
  7. package/dist/agent-profile-cell-B1V1gUZs.js.map +1 -0
  8. package/dist/{agent-profile-cell-BNp-Js9F.d.ts → agent-profile-cell-BcpFbNd5.d.ts} +2 -22
  9. package/dist/agent-profile-cell-BcpFbNd5.d.ts.map +1 -0
  10. package/dist/{agreement-irr-BEWLIF3_.js → agreement-irr-BhAqjluA.js} +5 -5
  11. package/dist/agreement-irr-BhAqjluA.js.map +1 -0
  12. package/dist/analyst/index.d.ts +36 -169
  13. package/dist/analyst/index.d.ts.map +1 -1
  14. package/dist/analyst/index.js +17 -192
  15. package/dist/analyst/index.js.map +1 -1
  16. package/dist/{analyze-runs-C84irm4i.js → analyze-runs-DqPJaJ4X.js} +225 -21
  17. package/dist/analyze-runs-DqPJaJ4X.js.map +1 -0
  18. package/dist/{analyze-runs-D68fd-su.d.ts → analyze-runs-DyeNcenW.d.ts} +5 -5
  19. package/dist/{analyze-runs-D68fd-su.d.ts.map → analyze-runs-DyeNcenW.d.ts.map} +1 -1
  20. package/dist/authenticity/index.d.ts +0 -23
  21. package/dist/authenticity/index.d.ts.map +1 -1
  22. package/dist/authenticity/index.js +1 -34
  23. package/dist/authenticity/index.js.map +1 -1
  24. package/dist/{benchmark-command-By0vh8q7.js → benchmark-command-xQK8nOI6.js} +554 -89
  25. package/dist/benchmark-command-xQK8nOI6.js.map +1 -0
  26. package/dist/benchmarks/index.d.ts +5 -6
  27. package/dist/benchmarks/index.d.ts.map +1 -1
  28. package/dist/benchmarks/index.js +4 -4
  29. package/dist/benchmarks/index.js.map +1 -1
  30. package/dist/builder-eval/index.d.ts +3 -3
  31. package/dist/builder-eval/index.js +2 -2
  32. package/dist/builder-eval/index.js.map +1 -1
  33. package/dist/campaign/index.d.ts +7 -7
  34. package/dist/campaign/index.js +15 -15
  35. package/dist/{campaign-BudZ70zo.js → campaign-CJGkxAQ6.js} +23 -530
  36. package/dist/campaign-CJGkxAQ6.js.map +1 -0
  37. package/dist/{campaign-evidence-Clrn4hBV.js → campaign-evidence-CQYw8_0H.js} +10 -18
  38. package/dist/{campaign-evidence-Clrn4hBV.js.map → campaign-evidence-CQYw8_0H.js.map} +1 -1
  39. package/dist/{canonical-B2RqLofv.js → canonical-Dq2hOQ8L.js} +2 -2
  40. package/dist/{canonical-B2RqLofv.js.map → canonical-Dq2hOQ8L.js.map} +1 -1
  41. package/dist/cell-spend-DEWR3nj-.js +90 -0
  42. package/dist/cell-spend-DEWR3nj-.js.map +1 -0
  43. package/dist/{chat-client-l-MVDlDk.js → chat-client-ldqxYYTN.js} +2 -2
  44. package/dist/{chat-client-l-MVDlDk.js.map → chat-client-ldqxYYTN.js.map} +1 -1
  45. package/dist/{chat-json-call-puMArmtW.js → chat-json-call-DavzexSp.js} +3 -2
  46. package/dist/chat-json-call-DavzexSp.js.map +1 -0
  47. package/dist/{claim-q4HQhK1G.js → claim-BPVuyHuy.js} +3 -3
  48. package/dist/{claim-q4HQhK1G.js.map → claim-BPVuyHuy.js.map} +1 -1
  49. package/dist/cli.js +113 -12
  50. package/dist/cli.js.map +1 -1
  51. package/dist/{client-BzL9niqi.d.ts → client-hwo3cJjR.d.ts} +3 -16
  52. package/dist/client-hwo3cJjR.d.ts.map +1 -0
  53. package/dist/{cluster-bootstrap-qcUWYFl-.js → cluster-bootstrap-DEwwhl0C.js} +3 -3
  54. package/dist/{cluster-bootstrap-qcUWYFl-.js.map → cluster-bootstrap-DEwwhl0C.js.map} +1 -1
  55. package/dist/{produced-state-BsDmnwxc.js → completion-verifier-B_jw6C9p.js} +9 -64
  56. package/dist/completion-verifier-B_jw6C9p.js.map +1 -0
  57. package/dist/contract/index.d.ts +12 -12
  58. package/dist/contract/index.js +25 -401
  59. package/dist/contract/index.js.map +1 -1
  60. package/dist/{cost-ledger-BP8V4kUQ.js → cost-ledger-D0352Z4d.js} +4 -8
  61. package/dist/cost-ledger-D0352Z4d.js.map +1 -0
  62. package/dist/dataset-DQqhOCPt.d.ts.map +1 -1
  63. package/dist/{default-registry-COhk9g36.js → default-registry-ClYjq3QE.js} +8 -11
  64. package/dist/default-registry-ClYjq3QE.js.map +1 -0
  65. package/dist/{default-registry-DzHiGruW.d.ts → default-registry-DXgLmoDI.d.ts} +7 -50
  66. package/dist/default-registry-DXgLmoDI.d.ts.map +1 -0
  67. package/dist/{define-agent-eval-CNTPzIkp.js → define-agent-eval-BDxNlP3D.js} +10 -10
  68. package/dist/{define-agent-eval-CNTPzIkp.js.map → define-agent-eval-BDxNlP3D.js.map} +1 -1
  69. package/dist/{define-agent-eval-WSck0CZJ.d.ts → define-agent-eval-BIkH_gAR.d.ts} +8 -8
  70. package/dist/define-agent-eval-BIkH_gAR.d.ts.map +1 -0
  71. package/dist/{descriptive-DboxPZrG.js → descriptive-CVNy48rR.js} +2 -2
  72. package/dist/{descriptive-DboxPZrG.js.map → descriptive-CVNy48rR.js.map} +1 -1
  73. package/dist/diagnosis/index.d.ts +4 -14
  74. package/dist/diagnosis/index.d.ts.map +1 -1
  75. package/dist/diagnosis/index.js +67 -28
  76. package/dist/diagnosis/index.js.map +1 -1
  77. package/dist/{dspy-rlm-engine-LRCu61m5.js → dspy-rlm-engine-DhdCr1Dc.js} +5 -10
  78. package/dist/dspy-rlm-engine-DhdCr1Dc.js.map +1 -0
  79. package/dist/{emitter-HcCmFQl_.d.ts → emitter-DABYYTcH.d.ts} +5 -5
  80. package/dist/emitter-DABYYTcH.d.ts.map +1 -0
  81. package/dist/emitter-DbMXUIVP.js.map +1 -1
  82. package/dist/{engine-3A_4jOlI.d.ts → engine-BvZ0R-36.d.ts} +17 -21
  83. package/dist/engine-BvZ0R-36.d.ts.map +1 -0
  84. package/dist/{equal-terms-BtdXayCJ.js → equal-terms-DbPGcVtI.js} +2 -2
  85. package/dist/equal-terms-DbPGcVtI.js.map +1 -0
  86. package/dist/{errors-DVZiqS7K.js → errors-BcOs5MKW.js} +2 -8
  87. package/dist/{errors-DVZiqS7K.js.map → errors-BcOs5MKW.js.map} +1 -1
  88. package/dist/{eval-campaign-90jeaCAI.js → eval-campaign-BLEfBGbA.js} +6 -6
  89. package/dist/{eval-campaign-90jeaCAI.js.map → eval-campaign-BLEfBGbA.js.map} +1 -1
  90. package/dist/{evaluation-6Yoi74gM.d.ts → evaluation-jTjC_QQD.d.ts} +7 -7
  91. package/dist/evaluation-jTjC_QQD.d.ts.map +1 -0
  92. package/dist/evaluation.d.ts +2 -2
  93. package/dist/evaluation.js +1 -1
  94. package/dist/evaluation.js.map +1 -1
  95. package/dist/{exact-types-ClugMJfI.d.ts → exact-types-CflbJ9f3.d.ts} +2 -2
  96. package/dist/{exact-types-ClugMJfI.d.ts.map → exact-types-CflbJ9f3.d.ts.map} +1 -1
  97. package/dist/{exec-ChJ6d5if.js → exec-CpbtOXVt.js} +2 -24
  98. package/dist/{exec-ChJ6d5if.js.map → exec-CpbtOXVt.js.map} +1 -1
  99. package/dist/execution-tracks-CpgFPpS5.js.map +1 -1
  100. package/dist/experiment/index.d.ts +9 -112
  101. package/dist/experiment/index.d.ts.map +1 -1
  102. package/dist/experiment/index.js +149 -241
  103. package/dist/experiment/index.js.map +1 -1
  104. package/dist/{exporters-X4xwla8A.js → exporters-Cl6IeNAu.js} +3 -3
  105. package/dist/exporters-Cl6IeNAu.js.map +1 -0
  106. package/dist/{exporters-ChTnR90l.d.ts → exporters-Dk3oEbK_.d.ts} +2 -4
  107. package/dist/{exporters-ChTnR90l.d.ts.map → exporters-Dk3oEbK_.d.ts.map} +1 -1
  108. package/dist/{external-optimizer-process-BjMGUuds.js → external-optimizer-process-KVFGvYX1.js} +4 -3
  109. package/dist/external-optimizer-process-KVFGvYX1.js.map +1 -0
  110. package/dist/{external-optimizer-subprocess-BqW0hHcI.js → external-optimizer-subprocess-XIHBOSKt.js} +9 -11
  111. package/dist/external-optimizer-subprocess-XIHBOSKt.js.map +1 -0
  112. package/dist/{extract-usage-BrnzRJ6H.js → extract-usage-BZLU01kR.js} +2 -23
  113. package/dist/{extract-usage-BrnzRJ6H.js.map → extract-usage-BZLU01kR.js.map} +1 -1
  114. package/dist/{failure-cluster-CbgEDcFJ.d.ts → failure-cluster-DN0t63Z0.d.ts} +4 -43
  115. package/dist/failure-cluster-DN0t63Z0.d.ts.map +1 -0
  116. package/dist/{failure-taxonomy-CjabTATj.js → failure-taxonomy-Bx5oWqaW.js} +2 -19
  117. package/dist/{failure-taxonomy-CjabTATj.js.map → failure-taxonomy-Bx5oWqaW.js.map} +1 -1
  118. package/dist/{feedback-trajectory-C4EG0nNH.d.ts → feedback-trajectory-BWmhRBGu.d.ts} +7 -9
  119. package/dist/feedback-trajectory-BWmhRBGu.d.ts.map +1 -0
  120. package/dist/{semantic-concept-judge-BQF8te0T.js → findings-store-BMVQtEsm.js} +9 -249
  121. package/dist/findings-store-BMVQtEsm.js.map +1 -0
  122. package/dist/fuzz.d.ts +19 -62
  123. package/dist/fuzz.d.ts.map +1 -1
  124. package/dist/fuzz.js +62 -108
  125. package/dist/fuzz.js.map +1 -1
  126. package/dist/{hf-dataset-BR2yhuNn.js → hf-dataset-D9f7-oX7.js} +4 -24
  127. package/dist/hf-dataset-D9f7-oX7.js.map +1 -0
  128. package/dist/hosted/index.d.ts +2 -2
  129. package/dist/hosted/index.js +2 -2
  130. package/dist/{index-DHKEAv3s.d.ts → index-9jJ2fKSX.d.ts} +2 -64
  131. package/dist/index-9jJ2fKSX.d.ts.map +1 -0
  132. package/dist/{index-DV-jSzxF.d.ts → index-BLGqZ81N.d.ts} +10 -15
  133. package/dist/{index-DV-jSzxF.d.ts.map → index-BLGqZ81N.d.ts.map} +1 -1
  134. package/dist/{index-DPcngtwt.d.ts → index-BiT6RtNH.d.ts} +11 -36
  135. package/dist/index-BiT6RtNH.d.ts.map +1 -0
  136. package/dist/{index-BO41eGj5.d.ts → index-CX-GAUpw.d.ts} +3 -7
  137. package/dist/index-CX-GAUpw.d.ts.map +1 -0
  138. package/dist/{index-BFRO6chs.d.ts → index-CiRwLv2R.d.ts} +5 -69
  139. package/dist/index-CiRwLv2R.d.ts.map +1 -0
  140. package/dist/{index-CyszSg4O.d.ts → index-ZuWx52WG.d.ts} +115 -347
  141. package/dist/index-ZuWx52WG.d.ts.map +1 -0
  142. package/dist/index.d.ts +211 -218
  143. package/dist/index.d.ts.map +1 -1
  144. package/dist/index.js +402 -1040
  145. package/dist/index.js.map +1 -1
  146. package/dist/{integrity-dLiDe2kD.js → integrity-DOd9Q86O.js} +7 -9
  147. package/dist/integrity-DOd9Q86O.js.map +1 -0
  148. package/dist/{integrity-BXOsAb3i.js → integrity-W4CADcFc.js} +3 -7
  149. package/dist/{integrity-BXOsAb3i.js.map → integrity-W4CADcFc.js.map} +1 -1
  150. package/dist/{integrity-DAMEDFaC.d.ts → integrity-_qgZLA5p.d.ts} +4 -6
  151. package/dist/{integrity-DAMEDFaC.d.ts.map → integrity-_qgZLA5p.d.ts.map} +1 -1
  152. package/dist/{internal-Epl-YBZW.js → internal-BTK91Bqt.js} +3 -3
  153. package/dist/{internal-Epl-YBZW.js.map → internal-BTK91Bqt.js.map} +1 -1
  154. package/dist/{interval-CELpye1n.js → interval-Dq9gt7jx.js} +3 -3
  155. package/dist/{interval-CELpye1n.js.map → interval-Dq9gt7jx.js.map} +1 -1
  156. package/dist/{journal-DfRz9kps.js → journal-CR4QikYp.js} +110 -110
  157. package/dist/journal-CR4QikYp.js.map +1 -0
  158. package/dist/json-BhdsJaOQ.js +111 -0
  159. package/dist/json-BhdsJaOQ.js.map +1 -0
  160. package/dist/{judge-calibration-BpCy5XQl.js → judge-calibration-BkXMIWA4.js} +3 -3
  161. package/dist/{judge-calibration-BpCy5XQl.js.map → judge-calibration-BkXMIWA4.js.map} +1 -1
  162. package/dist/{judge-gate-iY9s4L44.js → judge-gate-Co5tuh57.js} +4 -27
  163. package/dist/{judge-gate-iY9s4L44.js.map → judge-gate-Co5tuh57.js.map} +1 -1
  164. package/dist/{judge-gate-CE7CQAgt.d.ts → judge-gate-DsL0yFcb.d.ts} +8 -9
  165. package/dist/judge-gate-DsL0yFcb.d.ts.map +1 -0
  166. package/dist/{kind-factory-BobI5Jy3.js → kind-factory-Gy4fXixd.js} +88 -17
  167. package/dist/kind-factory-Gy4fXixd.js.map +1 -0
  168. package/dist/ledger-core/index.d.ts +2 -2
  169. package/dist/ledger-core/index.js +4 -4
  170. package/dist/{ledger-core-Gq7yp8lP.js → ledger-core-DrJhbms9.js} +5 -84
  171. package/dist/ledger-core-DrJhbms9.js.map +1 -0
  172. package/dist/{llm-client-CyBAoYiX.js → llm-client-BshCREn6.js} +6 -108
  173. package/dist/llm-client-BshCREn6.js.map +1 -0
  174. package/dist/{llm-judge-CG12SQJq.js → llm-judge-C8jMTNNG.js} +24 -32
  175. package/dist/llm-judge-C8jMTNNG.js.map +1 -0
  176. package/dist/matrix/index.d.ts +3 -3
  177. package/dist/matrix/index.js +3 -2
  178. package/dist/{matrix-Dm0BQW7U.js → matrix-B8l-Nsxd.js} +3 -88
  179. package/dist/matrix-B8l-Nsxd.js.map +1 -0
  180. package/dist/{matrix-DSrSxjjy.d.ts → matrix-DHe-dwVJ.d.ts} +9 -18
  181. package/dist/matrix-DHe-dwVJ.d.ts.map +1 -0
  182. package/dist/{mean-DHoMiudD.js → mean-BbA_JCzW.js} +39 -5
  183. package/dist/mean-BbA_JCzW.js.map +1 -0
  184. package/dist/meta-eval/index.d.ts +5 -5
  185. package/dist/meta-eval/index.js +10 -10
  186. package/dist/metrics-BKaPBtIG.js.map +1 -1
  187. package/dist/{mint-DNU8QPG4.d.ts → mint-BRZXuwRi.d.ts} +4 -4
  188. package/dist/{mint-DNU8QPG4.d.ts.map → mint-BRZXuwRi.d.ts.map} +1 -1
  189. package/dist/{mint-QaD5cA0u.js → mint-zszyYlpA.js} +4 -4
  190. package/dist/{mint-QaD5cA0u.js.map → mint-zszyYlpA.js.map} +1 -1
  191. package/dist/{multi-layer-verifier-BUaQ4C17.d.ts → multi-layer-verifier-DEgMUk6v.d.ts} +2 -2
  192. package/dist/{multi-layer-verifier-BUaQ4C17.d.ts.map → multi-layer-verifier-DEgMUk6v.d.ts.map} +1 -1
  193. package/dist/{multiplicity-DQUFcbW4.js → multiplicity-Sg0U7h8j.js} +2 -2
  194. package/dist/{multiplicity-DQUFcbW4.js.map → multiplicity-Sg0U7h8j.js.map} +1 -1
  195. package/dist/multishot/golden/index.d.ts +1 -23
  196. package/dist/multishot/golden/index.d.ts.map +1 -1
  197. package/dist/multishot/golden/index.js +3 -3
  198. package/dist/multishot/golden/index.js.map +1 -1
  199. package/dist/multishot/index.d.ts +3 -12
  200. package/dist/multishot/index.d.ts.map +1 -1
  201. package/dist/multishot/index.js +4 -3
  202. package/dist/multishot/index.js.map +1 -1
  203. package/dist/openapi.json +114 -10
  204. package/dist/{otel-spans-B9tUXRHF.js → otel-spans-CdN909k1.js} +3 -3
  205. package/dist/{otel-spans-B9tUXRHF.js.map → otel-spans-CdN909k1.js.map} +1 -1
  206. package/dist/{paired-arms-CwMGJsrS.js → paired-arms-Ct4LTKc1.js} +32 -9
  207. package/dist/paired-arms-Ct4LTKc1.js.map +1 -0
  208. package/dist/{paired-binary-CvpFHfKZ.js → paired-binary-B_nejA44.js} +3 -36
  209. package/dist/paired-binary-B_nejA44.js.map +1 -0
  210. package/dist/{paired-promotion-decision-DPKz8ZnF.d.ts → paired-promotion-decision-CQMJ5xGN.d.ts} +2 -34
  211. package/dist/paired-promotion-decision-CQMJ5xGN.d.ts.map +1 -0
  212. package/dist/{paired-promotion-decision-DsAPYGCm.js → paired-promotion-decision-D0SHb-ZL.js} +3 -3
  213. package/dist/{paired-promotion-decision-DsAPYGCm.js.map → paired-promotion-decision-D0SHb-ZL.js.map} +1 -1
  214. package/dist/{pareto-Cm6ndrWN.js → pareto-BzsxidFD.js} +11 -11
  215. package/dist/pareto-BzsxidFD.js.map +1 -0
  216. package/dist/pipelines/index.d.ts +16 -19
  217. package/dist/pipelines/index.d.ts.map +1 -1
  218. package/dist/pipelines/index.js +4 -68
  219. package/dist/pipelines/index.js.map +1 -1
  220. package/dist/{power-and-mde-Hj2ubb4B.js → power-and-mde-N5nU4_Ly.js} +3 -3
  221. package/dist/{power-and-mde-Hj2ubb4B.js.map → power-and-mde-N5nU4_Ly.js.map} +1 -1
  222. package/dist/{prime-protocol-CHKgA3a6.js → prime-protocol-BFgGrEMP.js} +3 -3
  223. package/dist/prime-protocol-BFgGrEMP.js.map +1 -0
  224. package/dist/{prime-protocol-HiB9HX7E.d.ts → prime-protocol-BWG2DQIk.d.ts} +3 -21
  225. package/dist/prime-protocol-BWG2DQIk.d.ts.map +1 -0
  226. package/dist/profile-cell.d.ts +2 -2
  227. package/dist/profile-cell.js +2 -2
  228. package/dist/{promotion-gate-BawrdPYr.d.ts → promotion-gate-DaY95yU6.d.ts} +7 -8
  229. package/dist/{promotion-gate-BawrdPYr.d.ts.map → promotion-gate-DaY95yU6.d.ts.map} +1 -1
  230. package/dist/{promotion-policy-KO57gbyV.js → promotion-policy-6A8osGr_.js} +4 -4
  231. package/dist/{promotion-policy-KO57gbyV.js.map → promotion-policy-6A8osGr_.js.map} +1 -1
  232. package/dist/{proposal-findings-CY1xKn_a.js → proposal-findings-BbWidvUk.js} +3 -3
  233. package/dist/proposal-findings-BbWidvUk.js.map +1 -0
  234. package/dist/{query-BUCZCK7F.js → query-B_RZ36an.js} +5 -5
  235. package/dist/query-B_RZ36an.js.map +1 -0
  236. package/dist/{query-DK1nWcmp.d.ts → query-Cmv_U8vf.d.ts} +4 -24
  237. package/dist/{query-DK1nWcmp.d.ts.map → query-Cmv_U8vf.d.ts.map} +1 -1
  238. package/dist/{random-ByUSnV9w.js → random-DDDcC7YU.js} +2 -2
  239. package/dist/{random-ByUSnV9w.js.map → random-DDDcC7YU.js.map} +1 -1
  240. package/dist/{raw-provider-sink-DoHz4JNG.js → raw-provider-sink-75w4FWBj.js} +2 -2
  241. package/dist/{raw-provider-sink-DoHz4JNG.js.map → raw-provider-sink-75w4FWBj.js.map} +1 -1
  242. package/dist/{redact-mtCCcT8z.d.ts → redact-DXdlFb_U.d.ts} +2 -3
  243. package/dist/redact-DXdlFb_U.d.ts.map +1 -0
  244. package/dist/{redact-B5xXKqQ3.js → redact-mtkTiJVk.js} +2 -7
  245. package/dist/{redact-B5xXKqQ3.js.map → redact-mtkTiJVk.js.map} +1 -1
  246. package/dist/{registry-D7_hm2fK.d.ts → registry-ssNoHzwD.d.ts} +4 -4
  247. package/dist/{registry-D7_hm2fK.d.ts.map → registry-ssNoHzwD.d.ts.map} +1 -1
  248. package/dist/{release-confidence-CpuP65S5.js → release-confidence-CYIFH-fp.js} +6 -11
  249. package/dist/release-confidence-CYIFH-fp.js.map +1 -0
  250. package/dist/{report-command-SC9nqb4M.js → render-C-w2EZ6p.js} +205 -363
  251. package/dist/render-C-w2EZ6p.js.map +1 -0
  252. package/dist/reporting.d.ts +6 -6
  253. package/dist/reporting.js +8 -8
  254. package/dist/{researcher-CEcfku-T.d.ts → researcher-0ABVIdGa.d.ts} +9 -9
  255. package/dist/{researcher-CEcfku-T.d.ts.map → researcher-0ABVIdGa.d.ts.map} +1 -1
  256. package/dist/{reward-CpCmAM8L.d.ts → reward-DrZ71vnf.d.ts} +3 -26
  257. package/dist/reward-DrZ71vnf.d.ts.map +1 -0
  258. package/dist/{reward-hacking-E0d-WVcP.js → reward-hacking-ByuETaIb.js} +3 -3
  259. package/dist/{reward-hacking-E0d-WVcP.js.map → reward-hacking-ByuETaIb.js.map} +1 -1
  260. package/dist/{reward-hacking-RVdZ6Kkt.d.ts → reward-hacking-E52wq86n.d.ts} +5 -18
  261. package/dist/reward-hacking-E52wq86n.d.ts.map +1 -0
  262. package/dist/reward-nw2xZGZG.js.map +1 -1
  263. package/dist/rl.d.ts +117 -295
  264. package/dist/rl.d.ts.map +1 -1
  265. package/dist/rl.js +27 -747
  266. package/dist/rl.js.map +1 -1
  267. package/dist/rollout/index.d.ts +5 -5
  268. package/dist/rollout/index.js +7 -8
  269. package/dist/{rollout-DZV1t2Wg.js → rollout-D6C50mfA.js} +88 -13
  270. package/dist/rollout-D6C50mfA.js.map +1 -0
  271. package/dist/{rubric-predictive-validity-DaFW70y9.js → rubric-predictive-validity-D3_oI72N.js} +3 -3
  272. package/dist/{rubric-predictive-validity-DaFW70y9.js.map → rubric-predictive-validity-D3_oI72N.js.map} +1 -1
  273. package/dist/{rubric-predictive-validity-C03EC7ZE.d.ts → rubric-predictive-validity-DiwT8PbL.d.ts} +2 -2
  274. package/dist/{rubric-predictive-validity-C03EC7ZE.d.ts.map → rubric-predictive-validity-DiwT8PbL.d.ts.map} +1 -1
  275. package/dist/{run-record-C1E24G2A.d.ts → run-record-CaxbLq6Y.d.ts} +4 -10
  276. package/dist/run-record-CaxbLq6Y.d.ts.map +1 -0
  277. package/dist/{run-record-C5e7FjhJ.js → run-record-CysKLB93.js} +3 -3
  278. package/dist/run-record-CysKLB93.js.map +1 -0
  279. package/dist/{run-record-Boc5IUTp.js → run-record-DhEJgoPz.js} +4 -4
  280. package/dist/{run-record-Boc5IUTp.js.map → run-record-DhEJgoPz.js.map} +1 -1
  281. package/dist/sandbox-harness-BlSOu4LX.d.ts.map +1 -1
  282. package/dist/{schema-CSf6qWgZ.js → schema-Bt2vLFm3.js} +2 -14
  283. package/dist/schema-Bt2vLFm3.js.map +1 -0
  284. package/dist/schema-C-ziyJD7.d.ts.map +1 -1
  285. package/dist/{schema-2g7trUwH.d.ts → schema-D6EVKzTB.d.ts} +4 -5
  286. package/dist/schema-D6EVKzTB.d.ts.map +1 -0
  287. package/dist/{schema-BuiujZ-g.js → schema-DxcKTzbi.js} +2 -2
  288. package/dist/schema-DxcKTzbi.js.map +1 -0
  289. package/dist/search/index.d.ts +3 -31
  290. package/dist/search/index.d.ts.map +1 -1
  291. package/dist/search/index.js +4 -4
  292. package/dist/{search-kernel-DyNEa6IU.d.ts → search-kernel-DxqjBxA-.d.ts} +10 -27
  293. package/dist/search-kernel-DxqjBxA-.d.ts.map +1 -0
  294. package/dist/{search-ledger-CVUGLDZp.js → search-ledger-TUvB5gWH.js} +7 -7
  295. package/dist/{search-ledger-CVUGLDZp.js.map → search-ledger-TUvB5gWH.js.map} +1 -1
  296. package/dist/{search-ledger-file-DDQUgvCg.js → search-ledger-file-UKCO2r8a.js} +4 -4
  297. package/dist/search-ledger-file-UKCO2r8a.js.map +1 -0
  298. package/dist/{search-shipper-BOSD46qv.js → search-shipper-Cw2z3-td.js} +4 -4
  299. package/dist/search-shipper-Cw2z3-td.js.map +1 -0
  300. package/dist/{search-summary-BQCSWf1g.js → search-summary-DQIvJsSB.js} +5 -5
  301. package/dist/{search-summary-BQCSWf1g.js.map → search-summary-DQIvJsSB.js.map} +1 -1
  302. package/dist/{sequential-CzK5DarL.js → sequential-BVv1sjsw.js} +2 -32
  303. package/dist/{sequential-CzK5DarL.js.map → sequential-BVv1sjsw.js.map} +1 -1
  304. package/dist/{sequential-Bf_pbE50.js → sequential-CimQrbjv.js} +5 -5
  305. package/dist/sequential-CimQrbjv.js.map +1 -0
  306. package/dist/{sequential-BhsrMupG.d.ts → sequential-KwGspfeR.d.ts} +2 -9
  307. package/dist/{sequential-BhsrMupG.d.ts.map → sequential-KwGspfeR.d.ts.map} +1 -1
  308. package/dist/{sequential-eprocess-DM_qENCl.js → sequential-eprocess-DodI6RoF.js} +2 -2
  309. package/dist/{sequential-eprocess-DM_qENCl.js.map → sequential-eprocess-DodI6RoF.js.map} +1 -1
  310. package/dist/{server-e9oWO_bN.js → server-DoVbqqtJ.js} +101 -101
  311. package/dist/server-DoVbqqtJ.js.map +1 -0
  312. package/dist/{single-run-lock-CV9Eyd_q.js → single-run-lock-U58yf8N6.js} +2 -2
  313. package/dist/{single-run-lock-CV9Eyd_q.js.map → single-run-lock-U58yf8N6.js.map} +1 -1
  314. package/dist/{skillopt-optimization-method-BiwA8nfX.js → skillopt-optimization-method-kyL94-ri.js} +17 -95
  315. package/dist/skillopt-optimization-method-kyL94-ri.js.map +1 -0
  316. package/dist/{spans-BHAUXMGN.d.ts → spans-CVngGij3.d.ts} +2 -2
  317. package/dist/{spans-BHAUXMGN.d.ts.map → spans-CVngGij3.d.ts.map} +1 -1
  318. package/dist/{statistical-heldout-42NlCNDe.js → statistical-heldout-Bksv92Th.js} +3 -3
  319. package/dist/statistical-heldout-Bksv92Th.js.map +1 -0
  320. package/dist/{statistical-heldout-CYpGaoDH.d.ts → statistical-heldout-C9AbgREN.d.ts} +8 -67
  321. package/dist/statistical-heldout-C9AbgREN.d.ts.map +1 -0
  322. package/dist/{steps-Bg5-KVx1.d.ts → steps-DlepJPwD.d.ts} +2 -56
  323. package/dist/steps-DlepJPwD.d.ts.map +1 -0
  324. package/dist/{store-C4avFjIi.d.ts → store-B3nWRy3B.d.ts} +2 -2
  325. package/dist/{store-C4avFjIi.d.ts.map → store-B3nWRy3B.d.ts.map} +1 -1
  326. package/dist/{store-otlp-DJ9dtA5x.js → store-otlp-BteDAbHK.js} +3 -3
  327. package/dist/{store-otlp-DJ9dtA5x.js.map → store-otlp-BteDAbHK.js.map} +1 -1
  328. package/dist/{store-tool-spans-CIX5arAG.js → store-tool-spans-D5l_OfTJ.js} +8 -8
  329. package/dist/store-tool-spans-D5l_OfTJ.js.map +1 -0
  330. package/dist/{store-tool-spans-DdCSPa7C.d.ts → store-tool-spans-DZS0HWVf.d.ts} +9 -18
  331. package/dist/store-tool-spans-DZS0HWVf.d.ts.map +1 -0
  332. package/dist/storyboard/index.d.ts +4 -11
  333. package/dist/storyboard/index.d.ts.map +1 -1
  334. package/dist/storyboard/index.js +1 -16
  335. package/dist/storyboard/index.js.map +1 -1
  336. package/dist/{summary-report-V_JACR4Y.js → summary-report-BsXJoYX5.js} +88 -10
  337. package/dist/summary-report-BsXJoYX5.js.map +1 -0
  338. package/dist/{summary-report-bdFDXq0b.d.ts → summary-report-DM7tlml1.d.ts} +10 -19
  339. package/dist/summary-report-DM7tlml1.d.ts.map +1 -0
  340. package/dist/supervisor-run/index.d.ts +60 -28
  341. package/dist/supervisor-run/index.d.ts.map +1 -1
  342. package/dist/supervisor-run/index.js +4 -458
  343. package/dist/supervisor-run/index.js.map +1 -1
  344. package/dist/{systemone-BomhHr2c.d.ts → systemone-Bp_Ac0p1.d.ts} +5 -5
  345. package/dist/{systemone-BomhHr2c.d.ts.map → systemone-Bp_Ac0p1.d.ts.map} +1 -1
  346. package/dist/systemone-protocol.js +1 -1
  347. package/dist/{systemone-review-CUVcQJu6.d.ts → systemone-review-CwE6JGIp.d.ts} +2 -2
  348. package/dist/{systemone-review-CUVcQJu6.d.ts.map → systemone-review-CwE6JGIp.d.ts.map} +1 -1
  349. package/dist/{systemone-review-D7sXwsqq.js → systemone-review-DzdOJ8I6.js} +3 -3
  350. package/dist/{systemone-review-D7sXwsqq.js.map → systemone-review-DzdOJ8I6.js.map} +1 -1
  351. package/dist/systemone.d.ts +5 -5
  352. package/dist/systemone.js +5 -5
  353. package/dist/systemone.js.map +1 -1
  354. package/dist/{task-failure-attributes-hEmASjix.js → task-failure-attributes-BMsblaDh.js} +3 -3
  355. package/dist/task-failure-attributes-BMsblaDh.js.map +1 -0
  356. package/dist/{terminal-record-DZyQZ3pz.js → terminal-record-78g6E4vh.js} +5 -5
  357. package/dist/terminal-record-78g6E4vh.js.map +1 -0
  358. package/dist/{tool-groups-C3NEcdCM.d.ts → tool-groups-DyNz2JZ_.d.ts} +3 -3
  359. package/dist/tool-groups-DyNz2JZ_.d.ts.map +1 -0
  360. package/dist/{tool-waste-BKgvd-TP.d.ts → tool-waste-CkjzWmeY.d.ts} +25 -28
  361. package/dist/tool-waste-CkjzWmeY.d.ts.map +1 -0
  362. package/dist/{tool-waste-BlesqTqW.js → tool-waste-DLsjS_p-.js} +5 -5
  363. package/dist/{tool-waste-BlesqTqW.js.map → tool-waste-DLsjS_p-.js.map} +1 -1
  364. package/dist/{trace-questions-CM2CAocU.d.ts → trace-questions-xxwgsuMi.d.ts} +4 -9
  365. package/dist/{trace-questions-CM2CAocU.d.ts.map → trace-questions-xxwgsuMi.d.ts.map} +1 -1
  366. package/dist/trace-repair/index.d.ts +14 -211
  367. package/dist/trace-repair/index.d.ts.map +1 -1
  368. package/dist/trace-repair/index.js +865 -981
  369. package/dist/trace-repair/index.js.map +1 -1
  370. package/dist/traces.d.ts +18 -46
  371. package/dist/traces.d.ts.map +1 -1
  372. package/dist/traces.js +13 -13
  373. package/dist/traces.js.map +1 -1
  374. package/dist/trajectory-replay/index.d.ts +3 -117
  375. package/dist/trajectory-replay/index.d.ts.map +1 -1
  376. package/dist/trajectory-replay/index.js +98 -56
  377. package/dist/trajectory-replay/index.js.map +1 -1
  378. package/dist/{tree-C1XYeXvy.js → tree-BGRglkKd.js} +13 -10
  379. package/dist/tree-BGRglkKd.js.map +1 -0
  380. package/dist/{types-B3jzCp0p.js → types-BXgASBCL.js} +2 -2
  381. package/dist/{types-B3jzCp0p.js.map → types-BXgASBCL.js.map} +1 -1
  382. package/dist/{types-C1tnkEgM.d.ts → types-CW8ewATZ.d.ts} +4 -113
  383. package/dist/types-CW8ewATZ.d.ts.map +1 -0
  384. package/dist/{types-BqkVmHfv.d.ts → types-CmXY0PQf.d.ts} +2 -4
  385. package/dist/types-CmXY0PQf.d.ts.map +1 -0
  386. package/dist/{types-C02qtnG8.d.ts → types-D-mgRzlh.d.ts} +7 -28
  387. package/dist/types-D-mgRzlh.d.ts.map +1 -0
  388. package/dist/{types-Cq8mSbq8.d.ts → types-D8GhJoHk.d.ts} +7 -9
  389. package/dist/{types-Cq8mSbq8.d.ts.map → types-D8GhJoHk.d.ts.map} +1 -1
  390. package/dist/{types-IXD2HJzu.js → types-DqFMNTgK.js} +2 -4
  391. package/dist/types-DqFMNTgK.js.map +1 -0
  392. package/dist/{types-Cbe2fWBa.d.ts → types-DtMbFw6U.d.ts} +3 -3
  393. package/dist/{types-Cbe2fWBa.d.ts.map → types-DtMbFw6U.d.ts.map} +1 -1
  394. package/dist/{verdict-Dl0tiT1X.js → verdict-BGQmxQ8A.js} +3 -3
  395. package/dist/verdict-BGQmxQ8A.js.map +1 -0
  396. package/dist/{verdict-E4eRNf7-.d.ts → verdict-C24NELtN.d.ts} +2 -15
  397. package/dist/{verdict-E4eRNf7-.d.ts.map → verdict-C24NELtN.d.ts.map} +1 -1
  398. package/dist/{verdict-cache-Bm8EKL2s.js → verdict-cache-CL_7tePD.js} +2 -2
  399. package/dist/verdict-cache-CL_7tePD.js.map +1 -0
  400. package/dist/wire/index.d.ts +31 -410
  401. package/dist/wire/index.d.ts.map +1 -1
  402. package/dist/wire/index.js +2 -2
  403. package/docs/changes/jev-evidence-review.md +1 -1
  404. package/docs/changes/jev-product-adoption.md +1 -1
  405. package/docs/changes/jev-recorded-evidence.md +1 -1
  406. package/docs/design/statistics-decisions.md +17 -17
  407. package/docs/design.md +1 -1
  408. package/docs/evaluation-integrity.md +1 -1
  409. package/docs/proofs/campaign-attempt-retention-20261001/README.md +2 -2
  410. package/docs/public-api.md +138 -529
  411. package/docs/rollout.md +1 -1
  412. package/docs/trace-repair-admission.md +1 -1
  413. package/docs/wire-protocol.md +1 -1
  414. package/package.json +1 -1
  415. package/dist/active-curriculum-B_Lc9kfO.js +0 -61
  416. package/dist/active-curriculum-B_Lc9kfO.js.map +0 -1
  417. package/dist/agent-profile-VFcQncbW.d.ts.map +0 -1
  418. package/dist/agent-profile-cell-BK8mlqlq.js.map +0 -1
  419. package/dist/agent-profile-cell-BNp-Js9F.d.ts.map +0 -1
  420. package/dist/agreement-irr-BEWLIF3_.js.map +0 -1
  421. package/dist/analyze-runs-C84irm4i.js.map +0 -1
  422. package/dist/baseline-Udi_PLuT.js +0 -259
  423. package/dist/baseline-Udi_PLuT.js.map +0 -1
  424. package/dist/benchmark-CXmVVHH2.js +0 -562
  425. package/dist/benchmark-CXmVVHH2.js.map +0 -1
  426. package/dist/benchmark-command-By0vh8q7.js.map +0 -1
  427. package/dist/campaign-BudZ70zo.js.map +0 -1
  428. package/dist/chat-json-call-puMArmtW.js.map +0 -1
  429. package/dist/claude-jsonl-dZir12m6.js +0 -86
  430. package/dist/claude-jsonl-dZir12m6.js.map +0 -1
  431. package/dist/client-BzL9niqi.d.ts.map +0 -1
  432. package/dist/cost-ledger-BP8V4kUQ.js.map +0 -1
  433. package/dist/counterfactual-CPaWN99-.d.ts +0 -60
  434. package/dist/counterfactual-CPaWN99-.d.ts.map +0 -1
  435. package/dist/counterfactual-Dn3veueY.js +0 -98
  436. package/dist/counterfactual-Dn3veueY.js.map +0 -1
  437. package/dist/default-registry-COhk9g36.js.map +0 -1
  438. package/dist/default-registry-DzHiGruW.d.ts.map +0 -1
  439. package/dist/define-agent-eval-WSck0CZJ.d.ts.map +0 -1
  440. package/dist/dspy-rlm-engine-LRCu61m5.js.map +0 -1
  441. package/dist/effect-sizes-BhzEpXMc.js +0 -82
  442. package/dist/effect-sizes-BhzEpXMc.js.map +0 -1
  443. package/dist/emitter-HcCmFQl_.d.ts.map +0 -1
  444. package/dist/engine-3A_4jOlI.d.ts.map +0 -1
  445. package/dist/equal-terms-BtdXayCJ.js.map +0 -1
  446. package/dist/evaluation-6Yoi74gM.d.ts.map +0 -1
  447. package/dist/exporters-X4xwla8A.js.map +0 -1
  448. package/dist/external-optimizer-process-BjMGUuds.js.map +0 -1
  449. package/dist/external-optimizer-subprocess-BqW0hHcI.js.map +0 -1
  450. package/dist/failure-cluster-CbgEDcFJ.d.ts.map +0 -1
  451. package/dist/feedback-trajectory-C4EG0nNH.d.ts.map +0 -1
  452. package/dist/hf-dataset-BR2yhuNn.js.map +0 -1
  453. package/dist/index-BFRO6chs.d.ts.map +0 -1
  454. package/dist/index-BO41eGj5.d.ts.map +0 -1
  455. package/dist/index-CyszSg4O.d.ts.map +0 -1
  456. package/dist/index-DHKEAv3s.d.ts.map +0 -1
  457. package/dist/index-DPcngtwt.d.ts.map +0 -1
  458. package/dist/integrity-dLiDe2kD.js.map +0 -1
  459. package/dist/journal-DfRz9kps.js.map +0 -1
  460. package/dist/judge-gate-CE7CQAgt.d.ts.map +0 -1
  461. package/dist/kind-factory-BobI5Jy3.js.map +0 -1
  462. package/dist/ledger-core-Gq7yp8lP.js.map +0 -1
  463. package/dist/llm-client-CyBAoYiX.js.map +0 -1
  464. package/dist/llm-judge-CG12SQJq.js.map +0 -1
  465. package/dist/matrix-DSrSxjjy.d.ts.map +0 -1
  466. package/dist/matrix-Dm0BQW7U.js.map +0 -1
  467. package/dist/mean-DHoMiudD.js.map +0 -1
  468. package/dist/normal-DsJZtrJc.js +0 -25
  469. package/dist/normal-DsJZtrJc.js.map +0 -1
  470. package/dist/paired-arms-CwMGJsrS.js.map +0 -1
  471. package/dist/paired-binary-CvpFHfKZ.js.map +0 -1
  472. package/dist/paired-promotion-decision-DPKz8ZnF.d.ts.map +0 -1
  473. package/dist/pareto-Cm6ndrWN.js.map +0 -1
  474. package/dist/prime-protocol-CHKgA3a6.js.map +0 -1
  475. package/dist/prime-protocol-HiB9HX7E.d.ts.map +0 -1
  476. package/dist/produced-state-BsDmnwxc.js.map +0 -1
  477. package/dist/proposal-findings-CY1xKn_a.js.map +0 -1
  478. package/dist/query-BUCZCK7F.js.map +0 -1
  479. package/dist/redact-mtCCcT8z.d.ts.map +0 -1
  480. package/dist/release-confidence-CpuP65S5.js.map +0 -1
  481. package/dist/report-command-SC9nqb4M.js.map +0 -1
  482. package/dist/reward-CpCmAM8L.d.ts.map +0 -1
  483. package/dist/reward-hacking-RVdZ6Kkt.d.ts.map +0 -1
  484. package/dist/rollout-DZV1t2Wg.js.map +0 -1
  485. package/dist/run-record-C1E24G2A.d.ts.map +0 -1
  486. package/dist/run-record-C5e7FjhJ.js.map +0 -1
  487. package/dist/schema-2g7trUwH.d.ts.map +0 -1
  488. package/dist/schema-BuiujZ-g.js.map +0 -1
  489. package/dist/schema-CSf6qWgZ.js.map +0 -1
  490. package/dist/search-kernel-DyNEa6IU.d.ts.map +0 -1
  491. package/dist/search-ledger-file-DDQUgvCg.js.map +0 -1
  492. package/dist/search-shipper-BOSD46qv.js.map +0 -1
  493. package/dist/semantic-concept-judge-BQF8te0T.js.map +0 -1
  494. package/dist/sequential-Bf_pbE50.js.map +0 -1
  495. package/dist/server-e9oWO_bN.js.map +0 -1
  496. package/dist/skillopt-optimization-method-BiwA8nfX.js.map +0 -1
  497. package/dist/statistical-heldout-42NlCNDe.js.map +0 -1
  498. package/dist/statistical-heldout-CYpGaoDH.d.ts.map +0 -1
  499. package/dist/steps-Bg5-KVx1.d.ts.map +0 -1
  500. package/dist/store-tool-spans-CIX5arAG.js.map +0 -1
  501. package/dist/store-tool-spans-DdCSPa7C.d.ts.map +0 -1
  502. package/dist/student-t-DVPW7CdL.js +0 -39
  503. package/dist/student-t-DVPW7CdL.js.map +0 -1
  504. package/dist/summary-report-V_JACR4Y.js.map +0 -1
  505. package/dist/summary-report-bdFDXq0b.d.ts.map +0 -1
  506. package/dist/task-failure-attributes-hEmASjix.js.map +0 -1
  507. package/dist/terminal-record-DZyQZ3pz.js.map +0 -1
  508. package/dist/tool-groups-C3NEcdCM.d.ts.map +0 -1
  509. package/dist/tool-waste-BKgvd-TP.d.ts.map +0 -1
  510. package/dist/trajectory-Y1fmdlw7.d.ts +0 -24
  511. package/dist/trajectory-Y1fmdlw7.d.ts.map +0 -1
  512. package/dist/tree-C1XYeXvy.js.map +0 -1
  513. package/dist/types-BqkVmHfv.d.ts.map +0 -1
  514. package/dist/types-C02qtnG8.d.ts.map +0 -1
  515. package/dist/types-C1tnkEgM.d.ts.map +0 -1
  516. package/dist/types-IXD2HJzu.js.map +0 -1
  517. package/dist/usage-receipt-Cx3xNeyB.js +0 -76
  518. package/dist/usage-receipt-Cx3xNeyB.js.map +0 -1
  519. package/dist/verdict-Dl0tiT1X.js.map +0 -1
  520. package/dist/verdict-cache-Bm8EKL2s.js.map +0 -1
  521. package/docs/design/self-improvement-evidence-audit.md +0 -263
package/dist/fuzz.d.ts CHANGED
@@ -1,5 +1,5 @@
1
1
  import { a as CostLedgerHandle, m as MaximumCharge, r as CostChannel } from "./cost-ledger-Jk62OBfF.js";
2
- import { t as DefaultVerdict } from "./verdict-E4eRNf7-.js";
2
+ import { t as DefaultVerdict } from "./verdict-C24NELtN.js";
3
3
  import { t as AdversarialMutation } from "./adversarial-smnADNFS.js";
4
4
  //#region src/fuzz/types.d.ts
5
5
  /** One input axis of the stratification plan (e.g. matterType, difficulty, personaRigor). */
@@ -39,13 +39,13 @@ interface Evaluation extends DefaultVerdict {
39
39
  /** Run the target against one scenario in a cell. */
40
40
  type Evaluator<S> = (scenario: S, cell: Cell) => Promise<Evaluation>;
41
41
  /** Context a proposer sees — prior elites + findings let a skill-backed proposer steer. */
42
- interface ProposeContext<S> {
42
+ interface FuzzProposeContext<S> {
43
43
  cell: Cell;
44
44
  seeds: S[];
45
45
  /** Current archive elites whose input cell matches. */
46
46
  elites: S[];
47
47
  /** Verified findings so far (read-only) — probe new gaps, not re-found ones. */
48
- findings: ReadonlyArray<Finding<S>>;
48
+ findings: ReadonlyArray<FuzzFinding<S>>;
49
49
  /** How many candidates to propose. */
50
50
  count: number;
51
51
  rng: () => number;
@@ -60,13 +60,13 @@ interface ProposeContext<S> {
60
60
  * is the scenario generator in the behavior-fuzzing loop. `SurfaceProposer` is
61
61
  * THE optimization proposer.
62
62
  */
63
- type MutationProposer<S> = (ctx: ProposeContext<S>) => Promise<S[]> | S[];
63
+ type MutationProposer<S> = (ctx: FuzzProposeContext<S>) => Promise<S[]> | S[];
64
64
  /**
65
65
  * What "interesting" means. `interest` in [0,1]; a candidate is notable (gate-
66
66
  * checked, reported) when `interest >= threshold`. `adversarialObjective` (low
67
67
  * score is interesting) and `noveltyObjective` (far from the archive) ship.
68
68
  */
69
- interface Objective {
69
+ interface FuzzObjective {
70
70
  kind: string;
71
71
  interest(ev: Evaluation, ctx: ObjectiveContext): number;
72
72
  /** Default 0.5. */
@@ -87,7 +87,7 @@ interface ValidityGates<S> {
87
87
  isUncontaminated?: (scenario: S, ev: Evaluation, cell: Cell) => boolean | Promise<boolean>;
88
88
  }
89
89
  /** A gate-verified, minimized finding — the unit the capsule reports. */
90
- interface Finding<S> {
90
+ interface FuzzFinding<S> {
91
91
  id: string;
92
92
  cell: Cell;
93
93
  scenario: S;
@@ -148,7 +148,7 @@ interface CapsuleData<S> {
148
148
  generatedAt?: string;
149
149
  coverage: CoverageCell[];
150
150
  /** Verified findings, sorted by descending interest. */
151
- findings: Finding<S>[];
151
+ findings: FuzzFinding<S>[];
152
152
  /** QD archive elites (binned by input × measured coords). */
153
153
  archive: ArchiveEntry<S>[];
154
154
  /** Post-harden lift, filled by a second pass after an improvement. */
@@ -211,7 +211,7 @@ type ExploreEvent<S> = {
211
211
  evaluation: Evaluation;
212
212
  } | {
213
213
  type: 'finding';
214
- finding: Finding<S>;
214
+ finding: FuzzFinding<S>;
215
215
  } | {
216
216
  type: 'eval-error';
217
217
  cell: Cell;
@@ -240,7 +240,7 @@ interface ExploreOptions<S> {
240
240
  /** Measured behavior coords appended to the archive bin. Default: input cell only. */
241
241
  descriptor?: (scenario: S, ev: Evaluation) => Record<string, string>;
242
242
  /** What "interesting" means. Default: `adversarialObjective(0.5)`. */
243
- objective?: Objective;
243
+ objective?: FuzzObjective;
244
244
  /** Validity gates. Default pass-through. */
245
245
  gates?: ValidityGates<S>;
246
246
  /** Budget steering across input cells. `variance` chases uncertainty; `uniform` is the unsteered ablation baseline. Default `variance`. */
@@ -322,7 +322,7 @@ interface BuildCapsuleInput<S> {
322
322
  /** The objective's notable threshold — drives findingRate. */
323
323
  threshold: number;
324
324
  archive: ArchiveEntry<S>[];
325
- findings: Finding<S>[];
325
+ findings: FuzzFinding<S>[];
326
326
  candidateFindings: number;
327
327
  runsUsed: number;
328
328
  /** Known-dollar / unknown-run split — present only when cost tracking was
@@ -339,7 +339,6 @@ interface BuildCapsuleInput<S> {
339
339
  detail: string;
340
340
  };
341
341
  }
342
- export declare function buildCapsule<S>(input: BuildCapsuleInput<S>): CapsuleData<S>;
343
342
  interface RenderCapsuleOptions {
344
343
  /** Max finding exemplars to show. Default 8. */
345
344
  maxFindings?: number;
@@ -349,47 +348,6 @@ interface RenderCapsuleOptions {
349
348
  /** Render a self-contained HTML capsule — heat-map + per-dimension chips + verified findings. */
350
349
  export declare function renderCapsuleHtml<S>(capsule: CapsuleData<S>, opts?: RenderCapsuleOptions): string;
351
350
  //#endregion
352
- //#region src/fuzz/explorer.d.ts
353
- export declare class BehaviorExplorer<S> {
354
- private readonly opts;
355
- private readonly cells;
356
- private readonly cellById;
357
- private readonly objective;
358
- private readonly threshold;
359
- private readonly floorPerCell;
360
- private readonly perRoundBudget;
361
- /** The single evaluation log — coverage + allocation are projections of it. */
362
- private readonly log;
363
- /** binId (input × measured coords) → the most interesting entry seen. */
364
- private readonly archiveByBin;
365
- private readonly _findings;
366
- private runsUsed;
367
- private candidateFindings;
368
- private evalErrors;
369
- private consecutiveEvalErrors;
370
- private stoppedEarly;
371
- private readonly rng;
372
- private readonly costLedger?;
373
- private readonly costPhase;
374
- constructor(opts: ExploreOptions<S>);
375
- private binId;
376
- private allocate;
377
- private objectiveContext;
378
- /** Elites whose INPUT cell matches — what the proposer mutates/deepens from. */
379
- private elitesFor;
380
- /** One allocate → propose → evaluate → gate → archive round. */
381
- step(): Promise<{
382
- runs: number;
383
- findings: Finding<S>[];
384
- }>;
385
- /** Loop `step()` until the run or dollar budget is spent, the signal aborts,
386
- * or no progress is made. */
387
- run(): Promise<CapsuleData<S>>;
388
- coverage(): CoverageCell[];
389
- findings(): Finding<S>[];
390
- capsule(): CapsuleData<S>;
391
- }
392
- //#endregion
393
351
  //#region src/fuzz/fuzz-agent.d.ts
394
352
  type FuzzAgentOptions<S> = Omit<ExploreOptions<S>, 'objective'> & {
395
353
  /** Score strictly below this is a candidate failure. Default 0.5. */
@@ -435,16 +393,16 @@ export declare function mutationProposer<S>(opts: {
435
393
  mutationsFor: (cell: Cell) => AdversarialMutation<S>[];
436
394
  scenarioId: (s: S) => string;
437
395
  }): MutationProposer<S>;
438
- /** Adversarial: a low headline score is interesting — find where the agent fails. */
439
- export declare function adversarialObjective(threshold?: number): Objective;
440
- /**
441
- * Novelty: interesting when far from the archive in score AND measured behavior
442
- * descriptor — quality-diversity's diversity pressure; drives corpus growth
443
- * rather than re-finding the same hole.
444
- */
445
- export declare function noveltyObjective(threshold?: number): Objective;
446
396
  //#endregion
447
397
  //#region src/fuzz/tools.d.ts
398
+ /**
399
+ * Agent-drivable surface over a live exploration session.
400
+ *
401
+ * Framework-neutral tool defs ({name, description, parameters: JSON Schema,
402
+ * handler}) so the on-demand agent — not a batch script — drives the search:
403
+ * step it, read coverage, inspect findings, render the capsule. Transport
404
+ * encodings (OpenAI function shape, MCP) are one-line mappings the host owns.
405
+ */
448
406
  interface ExploreToolDef {
449
407
  name: string;
450
408
  description: string;
@@ -454,7 +412,6 @@ interface ExploreToolDef {
454
412
  signal?: AbortSignal;
455
413
  }) => Promise<unknown>;
456
414
  }
457
- export declare function makeExploreTools<S>(explorer: BehaviorExplorer<S>): ExploreToolDef[];
458
415
  //#endregion
459
- export type { ArchiveEntry, BehaviorSpace, BuildCapsuleInput, CapsuleData, Cell, CoverageCell, EvalRecord, Evaluation, Evaluator, ExploreEvent, ExploreOptions, ExploreToolDef, Finding, FuzzAgentOptions, MutationProposer, Objective, ObjectiveContext, ProposeContext, RenderCapsuleOptions, RunCost, SpaceAxis, ValidityGates };
416
+ export type { ArchiveEntry, BehaviorSpace, BuildCapsuleInput, CapsuleData, Cell, CoverageCell, EvalRecord, Evaluation, Evaluator, ExploreEvent, ExploreOptions, ExploreToolDef, FuzzAgentOptions, FuzzFinding, FuzzObjective, FuzzProposeContext, MutationProposer, ObjectiveContext, RenderCapsuleOptions, RunCost, SpaceAxis, ValidityGates };
460
417
  //# sourceMappingURL=fuzz.d.ts.map
@@ -1 +1 @@
1
- {"version":3,"file":"fuzz.d.ts","names":[],"sources":["../src/fuzz/types.ts","../src/fuzz/cube.ts","../src/fuzz/capsule.ts","../src/fuzz/explorer.ts","../src/fuzz/fuzz-agent.ts","../src/fuzz/gates.ts","../src/fuzz/policies.ts","../src/fuzz/tools.ts"],"mappings":";;;;;UA2BiB;EACf;EACA;;;UAIe;EACf,MAAM;;;UAIS;;EAEf;EACA,QAAQ;;;;;;;;UAWO,mBAAmB;;EAElC,aAAa;;EAEb;;EAEA;;EAEA;;;EAGA;;;KAIU,UAAU,MAAM,UAAU,GAAG,MAAM,SAAS,QAAQ;;UAK/C,eAAe;EAC9B,MAAM;EACN,OAAO;;EAEP,QAAQ;;EAER,UAAU,cAAc,QAAQ;;EAEhC;EACA;;;;;;;;;;;;KAaU,iBAAiB,MAAM,KAAK,eAAe,OAAO,QAAQ,OAAO;;;;;;UAO5D;EACf;EACA,SAAS,IAAI,YAAY,KAAK;;EAE9B;;UAGe;EACf;EACA,oBAAoB,MAAM;;;;;;;;UASX,cAAc;EAC7B,WAAW,UAAU,GAAG,IAAI,YAAY,MAAM,mBAAmB;EACjE,oBAAoB,UAAU,GAAG,IAAI,YAAY,MAAM,mBAAmB;;;UAM3D,QAAQ;EACvB;EACA,MAAM;EACN,UAAU;;EAEV,WAAW;;EAEX;;EAEA,YAAY;;EAEZ;;EAEA;;;UAIe,aAAa;;EAE5B;EACA,MAAM;EACN,UAAU;EACV,YAAY;EACZ;;;UAIe;EACf;EACA;EACA;EACA;EACA;EACA;;;UAIe;EACf,MAAM;EACN;;;EAGA,OAAO;;EAEP;;;EAGA,YAAY,eAAe;;;EAG3B,WAAW;;;EAGX;EACA;;;UAIe,YAAY;EAC3B;EACA;;EAEA;EACA,UAAU;;EAEV,UAAU,QAAQ;;EAElB,SAAS,aAAa;;EAEtB;IAAS;IAAgB;IAAe;;EACxC;IACE;;IAEA;IACA;;IAEA;IACA;IACA;;;;IAIA,YAAY;;IAEZ,WAAW;;;IAGX;;;IAGA;;;;IAIA;;;;IAIA;MAAiB;MAAuB;;;;;;;;;UAW3B;EACf;EACA;;KAKU,aAAa;EACnB;EAAwB,MAAM;EAAM;;EACpC;EAAmB,MAAM;EAAM,UAAU;EAAG,YAAY;;EACxD;EAAiB,SAAS,QAAQ;;EAClC;EAAoB,MAAM;EAAM;EAAoB;;EACpD;EAAe;EAAkB;;UAEtB,eAAe;;EAE9B;;EAEA,OAAO;;EAEP,UAAU,iBAAiB;;EAE3B,UAAU,UAAU;;EAEpB,WAAW,MAAM,SAAS,MAAM,QAAQ;;EAExC,aAAa,UAAU;;EAEvB,gBAAgB,UAAU;;EAE1B,cAAc,UAAU,GAAG,IAAI,eAAe;;EAE9C,YAAY;;EAEZ,QAAQ,cAAc;;EAEtB;;EAEA;;EAEA;;EAEA,YAAY,UAAU,GAAG,UAAU,UAAU,IAAI,MAAM,SAAS,QAAQ,KAAK;;EAE7E;;;EAGA;;EAEA,SAAS;;EAET,cAAc,OAAO,aAAa;;EAElC;;;;;;;EAOA,UAAU,UAAU,GAAG,MAAM,MAAM,IAAI,eAAe;;EAEtD,mBAAmB,UAAU,GAAG,MAAM,SAAS;;;;;EAK/C;;;;;;;EAOA,SAAS;;EAET,UAAU;IAAS;IAAa,SAAS;;;;;;UC1S1B;EACf,MAAM;EACN,IAAI;;EAEJ;;EAEA;;;EAGA;;;wBAIc,eAAe,OAAO,gBAAgB;;wBActC,OAAO,OAAO,eAAe,QAAQ;;;;;wBA+BrC,cAAc,OAAO,QAAQ,KAAK,cAAc,oBAAoB;;;UC1DnE,kBAAkB;EACjC;EACA;EACA,OAAO;EACP,KAAK;;EAEL;EACA,SAAS,aAAa;EACtB,UAAU,QAAQ;EAClB;EACA;;;EAGA;IAAS;IAAiB;;;EAE1B;;EAEA;IAAiB;IAAuB;;;wBAG1B,aAAa,GAAG,OAAO,kBAAkB,KAAK,YAAY;UAsIzD;;EAEf;;EAEA;;;wBAIc,kBAAkB,GAChC,SAAS,YAAY,IACrB,OAAM;;;qBC5HK,iBAAiB;mBAsBC;mBArBZ;mBACA;mBACA;mBACA;mBACA;mBACA;;mBAGA;;mBAEA;mBACA;UACT;UACA;UACA;UACA;UACA;mBACS;mBACA;mBACA;EAEjB,YAA6B,MAAM,eAAe;UAkD1C;UASA;UAgBA;;UASA;;EAOF,QAAQ;IAAU;IAAc,UAAU,QAAQ;;;;EAyLlD,OAAO,QAAQ,YAAY;EAYjC,YAAY;EAIZ,YAAY,QAAQ;EAIpB,WAAW,YAAY;;;;KCxWb,iBAAiB,KAAK,KAAK,eAAe;;EAEpD;;wBAGoB,UAAU,GAC9B,MAAM,iBAAiB,KACtB;EAAU,SAAS,YAAY;;;;;wBCRlB,aAAa,MAAM,MAAM,MAAM,cAAc,kBAAkB,cAAc;;;;;;;;wBAyB7E,0BAA0B,GAAG;EAC3C,UAAU,UAAU;;EAEpB,UAAU,UAAU,MAAM;;EAE1B;IACE,cAAc;;;;;wBAgBF,kBAAkB,GAAG;EACnC;EACA;IACE,cAAc;;;;;;;;wBCjCF,iBAAiB,GAAG;EAClC,eAAe,MAAM,SAAS,oBAAoB;EAClD,aAAa,GAAG;IACd,iBAAiB;;wBA2BL,qBAAqB,qBAAkB;;;;;;wBAqBvC,iBAAiB,qBAAkB;;;UCnElC;EACf;EACA;;EAEA,YAAY;EACZ,UAAU,eAAe;IAAQ,SAAS;QAAkB;;wBAK9C,iBAAiB,GAAG,UAAU,iBAAiB,KAAK"}
1
+ {"version":3,"file":"fuzz.d.ts","names":[],"sources":["../src/fuzz/types.ts","../src/fuzz/cube.ts","../src/fuzz/capsule.ts","../src/fuzz/fuzz-agent.ts","../src/fuzz/gates.ts","../src/fuzz/policies.ts","../src/fuzz/tools.ts"],"mappings":";;;;;UA2BiB;EACf;EACA;;;UAIe;EACf,MAAM;;;UAIS;;EAEf;EACA,QAAQ;;;;;;;;UAWO,mBAAmB;;EAElC,aAAa;;EAEb;;EAEA;;EAEA;;;EAGA;;;KAIU,UAAU,MAAM,UAAU,GAAG,MAAM,SAAS,QAAQ;;UAK/C,mBAAmB;EAClC,MAAM;EACN,OAAO;;EAEP,QAAQ;;EAER,UAAU,cAAc,YAAY;;EAEpC;EACA;;;;;;;;;;;;KAaU,iBAAiB,MAAM,KAAK,mBAAmB,OAAO,QAAQ,OAAO;;;;;;UAOhE;EACf;EACA,SAAS,IAAI,YAAY,KAAK;;EAE9B;;UAGe;EACf;EACA,oBAAoB,MAAM;;;;;;;;UASX,cAAc;EAC7B,WAAW,UAAU,GAAG,IAAI,YAAY,MAAM,mBAAmB;EACjE,oBAAoB,UAAU,GAAG,IAAI,YAAY,MAAM,mBAAmB;;;UAM3D,YAAY;EAC3B;EACA,MAAM;EACN,UAAU;;EAEV,WAAW;;EAEX;;EAEA,YAAY;;EAEZ;;EAEA;;;UAIe,aAAa;;EAE5B;EACA,MAAM;EACN,UAAU;EACV,YAAY;EACZ;;;UAIe;EACf;EACA;EACA;EACA;EACA;EACA;;;UAIe;EACf,MAAM;EACN;;;EAGA,OAAO;;EAEP;;;EAGA,YAAY,eAAe;;;EAG3B,WAAW;;;EAGX;EACA;;;UAIe,YAAY;EAC3B;EACA;;EAEA;EACA,UAAU;;EAEV,UAAU,YAAY;;EAEtB,SAAS,aAAa;;EAEtB;IAAS;IAAgB;IAAe;;EACxC;IACE;;IAEA;IACA;;IAEA;IACA;IACA;;;;IAIA,YAAY;;IAEZ,WAAW;;;IAGX;;;IAGA;;;;IAIA;;;;IAIA;MAAiB;MAAuB;;;;;;;;;UAW3B;EACf;EACA;;KAKU,aAAa;EACnB;EAAwB,MAAM;EAAM;;EACpC;EAAmB,MAAM;EAAM,UAAU;EAAG,YAAY;;EACxD;EAAiB,SAAS,YAAY;;EACtC;EAAoB,MAAM;EAAM;EAAoB;;EACpD;EAAe;EAAkB;;UAEtB,eAAe;;EAE9B;;EAEA,OAAO;;EAEP,UAAU,iBAAiB;;EAE3B,UAAU,UAAU;;EAEpB,WAAW,MAAM,SAAS,MAAM,QAAQ;;EAExC,aAAa,UAAU;;EAEvB,gBAAgB,UAAU;;EAE1B,cAAc,UAAU,GAAG,IAAI,eAAe;;EAE9C,YAAY;;EAEZ,QAAQ,cAAc;;EAEtB;;EAEA;;EAEA;;EAEA,YAAY,UAAU,GAAG,UAAU,UAAU,IAAI,MAAM,SAAS,QAAQ,KAAK;;EAE7E;;;EAGA;;EAEA,SAAS;;EAET,cAAc,OAAO,aAAa;;EAElC;;;;;;;EAOA,UAAU,UAAU,GAAG,MAAM,MAAM,IAAI,eAAe;;EAEtD,mBAAmB,UAAU,GAAG,MAAM,SAAS;;;;;EAK/C;;;;;;;EAOA,SAAS;;EAET,UAAU;IAAS;IAAa,SAAS;;;;;;UC1S1B;EACf,MAAM;EACN,IAAI;;EAEJ;;EAEA;;;EAGA;;;wBAIc,eAAe,OAAO,gBAAgB;;wBActC,OAAO,OAAO,eAAe,QAAQ;;;;;wBA+BrC,cAAc,OAAO,QAAQ,KAAK,cAAc,oBAAoB;;;UC1DnE,kBAAkB;EACjC;EACA;EACA,OAAO;EACP,KAAK;;EAEL;EACA,SAAS,aAAa;EACtB,UAAU,YAAY;EACtB;EACA;;;EAGA;IAAS;IAAiB;;;EAE1B;;EAEA;IAAiB;IAAuB;;;UAyIzB;;EAEf;;EAEA;;;wBAIc,kBAAkB,GAChC,SAAS,YAAY,IACrB,OAAM;;;KCtKI,iBAAiB,KAAK,KAAK,eAAe;;EAEpD;;wBAGoB,UAAU,GAC9B,MAAM,iBAAiB,KACtB;EAAU,SAAS,YAAY;;;;;wBCRlB,aAAa,MAAM,MAAM,MAAM,cAAc,kBAAkB,cAAc;;;;;;;;wBAyB7E,0BAA0B,GAAG;EAC3C,UAAU,UAAU;;EAEpB,UAAU,UAAU,MAAM;;EAE1B;IACE,cAAc;;;;;wBAgBF,kBAAkB,GAAG;EACnC;EACA;IACE,cAAc;;;;;;;;wBCnCF,iBAAiB,GAAG;EAClC,eAAe,MAAM,SAAS,oBAAoB;EAClD,aAAa,GAAG;IACd,iBAAiB;;;;;;;;;;;UCjBJ;EACf;EACA;;EAEA,YAAY;EACZ,UAAU,eAAe;IAAQ,SAAS;QAAkB"}
package/dist/fuzz.js CHANGED
@@ -1,7 +1,6 @@
1
- import { s as ValidationError } from "./errors-DVZiqS7K.js";
2
- import { t as mulberry32 } from "./random-ByUSnV9w.js";
3
- import { o as CostReceiptCaptureError, t as CostLedger } from "./cost-ledger-BP8V4kUQ.js";
4
- import { t as varianceBasedCurriculum } from "./active-curriculum-B_Lc9kfO.js";
1
+ import { s as ValidationError } from "./errors-BcOs5MKW.js";
2
+ import { t as mulberry32 } from "./random-DDDcC7YU.js";
3
+ import { o as CostReceiptCaptureError, t as CostLedger } from "./cost-ledger-D0352Z4d.js";
5
4
  //#region src/fuzz/cube.ts
6
5
  /** Enumerate every input cell (cartesian product of the axes), in stable order. */
7
6
  function enumerateCells(space) {
@@ -260,6 +259,64 @@ ${findingsHtml(capsule.findings, opts.maxFindings ?? 8)}
260
259
  </div></body></html>`;
261
260
  }
262
261
  //#endregion
262
+ //#region src/rl/active-curriculum.ts
263
+ /**
264
+ * Variance-proportional allocation. For each cell, estimate variance from
265
+ * past observations + a prior, then allocate the budget proportional to
266
+ * (sqrt(variance) + 1/sqrt(n)) — a classical optimal-allocation rule
267
+ * (Neyman 1934) that balances "explore noisy cells" with "explore
268
+ * under-sampled cells."
269
+ */
270
+ function varianceBasedCurriculum(observations, candidateCells, opts) {
271
+ const variancePrior = opts.variancePrior ?? .05;
272
+ const floor = opts.floorPerCell ?? 1;
273
+ const budget = opts.budget;
274
+ const grouped = /* @__PURE__ */ new Map();
275
+ for (const o of observations) {
276
+ const k = `${o.variantId}::${o.scenarioId}`;
277
+ const arr = grouped.get(k) ?? [];
278
+ arr.push(o.score);
279
+ grouped.set(k, arr);
280
+ }
281
+ const cellStats = candidateCells.map((c) => {
282
+ const k = `${c.variantId}::${c.scenarioId}`;
283
+ const samples = grouped.get(k) ?? [];
284
+ const n = samples.length;
285
+ const mean = n === 0 ? .5 : samples.reduce((s, v) => s + v, 0) / n;
286
+ const variance = n < 2 ? variancePrior : samples.reduce((s, v) => s + (v - mean) ** 2, 0) / (n - 1) + variancePrior;
287
+ const weight = Math.sqrt(variance) + 1 / Math.sqrt(Math.max(1, n));
288
+ return {
289
+ variantId: c.variantId,
290
+ scenarioId: c.scenarioId,
291
+ n,
292
+ mean,
293
+ variance,
294
+ weight
295
+ };
296
+ });
297
+ const floorTotal = floor * cellStats.length;
298
+ if (floorTotal >= budget) {
299
+ const each = Math.max(1, Math.floor(budget / Math.max(1, cellStats.length)));
300
+ return cellStats.map((c) => ({
301
+ variantId: c.variantId,
302
+ scenarioId: c.scenarioId,
303
+ count: each,
304
+ reason: `floor allocation (budget tight; n=${c.n})`
305
+ }));
306
+ }
307
+ const remaining = budget - floorTotal;
308
+ const totalWeight = cellStats.reduce((s, c) => s + c.weight, 0);
309
+ return cellStats.map((c) => {
310
+ const proportional = totalWeight === 0 ? 0 : Math.round(c.weight / totalWeight * remaining);
311
+ return {
312
+ variantId: c.variantId,
313
+ scenarioId: c.scenarioId,
314
+ count: floor + proportional,
315
+ reason: `variance ${c.variance.toFixed(3)} (n=${c.n}, mean=${c.mean.toFixed(3)})`
316
+ };
317
+ });
318
+ }
319
+ //#endregion
263
320
  //#region src/fuzz/policies.ts
264
321
  const clamp01 = (x) => Math.max(0, Math.min(1, x));
265
322
  /**
@@ -298,30 +355,6 @@ function adversarialObjective(threshold = .5) {
298
355
  interest: (ev) => clamp01(1 - ev.score)
299
356
  };
300
357
  }
301
- function hamming(a, b) {
302
- if (!a || !b) return 1;
303
- const keys = /* @__PURE__ */ new Set([...Object.keys(a), ...Object.keys(b)]);
304
- if (keys.size === 0) return 0;
305
- let diff = 0;
306
- for (const k of keys) if (a[k] !== b[k]) diff++;
307
- return diff / keys.size;
308
- }
309
- /**
310
- * Novelty: interesting when far from the archive in score AND measured behavior
311
- * descriptor — quality-diversity's diversity pressure; drives corpus growth
312
- * rather than re-finding the same hole.
313
- */
314
- function noveltyObjective(threshold = .3) {
315
- return {
316
- kind: "novelty",
317
- threshold,
318
- interest: (ev, ctx) => {
319
- const scoreNovelty = ctx.archiveScores.length === 0 ? 1 : Math.min(...ctx.archiveScores.map((s) => Math.abs(s - ev.score)));
320
- const descNovelty = ctx.archiveDescriptors.length === 0 ? 1 : Math.min(...ctx.archiveDescriptors.map((d) => hamming(d, ev.descriptor)));
321
- return clamp01(.5 * scoreNovelty + .5 * descNovelty);
322
- }
323
- };
324
- }
325
358
  //#endregion
326
359
  //#region src/fuzz/explorer.ts
327
360
  /**
@@ -677,85 +710,6 @@ function severityFloorGate(opts) {
677
710
  return { isValid: (_scenario, ev) => ev.score <= threshold - margin };
678
711
  }
679
712
  //#endregion
680
- //#region src/fuzz/tools.ts
681
- /**
682
- * Agent-drivable surface over a live exploration session.
683
- *
684
- * Framework-neutral tool defs ({name, description, parameters: JSON Schema,
685
- * handler}) so the on-demand agent — not a batch script — drives the search:
686
- * step it, read coverage, inspect findings, render the capsule. Transport
687
- * encodings (OpenAI function shape, MCP) are one-line mappings the host owns.
688
- */
689
- const NO_ARGS = {
690
- type: "object",
691
- properties: {},
692
- additionalProperties: false
693
- };
694
- function makeExploreTools(explorer) {
695
- return [
696
- {
697
- name: "explore_step",
698
- description: "Run one exploration round: allocate budget across cells, propose + evaluate scenarios, archive elites, admit gate-verified findings. Returns runs spent and new findings.",
699
- parameters: NO_ARGS,
700
- handler: async () => {
701
- const { runs, findings } = await explorer.step();
702
- return {
703
- runs,
704
- newFindings: findings.length,
705
- findings
706
- };
707
- }
708
- },
709
- {
710
- name: "explore_coverage",
711
- description: "Read the live coverage map: per planned cell — runs, robustness (null = uncovered), finding rate, per-dimension means.",
712
- parameters: NO_ARGS,
713
- handler: async () => explorer.coverage()
714
- },
715
- {
716
- name: "explore_findings",
717
- description: "List gate-verified findings so far, sorted by descending interest.",
718
- parameters: {
719
- type: "object",
720
- properties: { limit: {
721
- type: "number",
722
- description: "max findings to return"
723
- } },
724
- additionalProperties: false
725
- },
726
- handler: async (args) => {
727
- const limit = args?.limit;
728
- const all = explorer.findings();
729
- return typeof limit === "number" ? all.slice(0, limit) : all;
730
- }
731
- },
732
- {
733
- name: "explore_capsule",
734
- description: "Build the capsule artifact from the current session state. format \"data\" returns the structured CapsuleData; \"html\" returns the standalone page.",
735
- parameters: {
736
- type: "object",
737
- properties: {
738
- format: {
739
- type: "string",
740
- enum: ["data", "html"]
741
- },
742
- generatedAt: {
743
- type: "string",
744
- description: "ISO timestamp to stamp into the page"
745
- }
746
- },
747
- additionalProperties: false
748
- },
749
- handler: async (args) => {
750
- const a = args ?? {};
751
- const capsule = explorer.capsule();
752
- if (a.format === "html") return renderCapsuleHtml(capsule, { generatedAt: a.generatedAt });
753
- return capsule;
754
- }
755
- }
756
- ];
757
- }
758
- //#endregion
759
- export { BehaviorExplorer, adversarialObjective, buildCapsule, buildCoverage, cellId, composeGates, enumerateCells, fuzzAgent, makeExploreTools, mutationProposer, noveltyObjective, perturbationStabilityGate, renderCapsuleHtml, severityFloorGate };
713
+ export { buildCoverage, cellId, composeGates, enumerateCells, fuzzAgent, mutationProposer, perturbationStabilityGate, renderCapsuleHtml, severityFloorGate };
760
714
 
761
715
  //# sourceMappingURL=fuzz.js.map