@tangle-network/agent-eval 0.120.0 → 0.120.2

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 (171) hide show
  1. package/CHANGELOG.md +12 -0
  2. package/package.json +1 -1
  3. package/dist/analyst/index.d.ts +0 -3111
  4. package/dist/analyst/index.js +0 -403
  5. package/dist/analyst/index.js.map +0 -1
  6. package/dist/authenticity/index.d.ts +0 -161
  7. package/dist/authenticity/index.js +0 -215
  8. package/dist/authenticity/index.js.map +0 -1
  9. package/dist/belief-state/index.d.ts +0 -1301
  10. package/dist/belief-state/index.js +0 -2152
  11. package/dist/belief-state/index.js.map +0 -1
  12. package/dist/benchmarks/index.d.ts +0 -974
  13. package/dist/benchmarks/index.js +0 -60
  14. package/dist/benchmarks/index.js.map +0 -1
  15. package/dist/builder-eval/index.d.ts +0 -695
  16. package/dist/builder-eval/index.js +0 -366
  17. package/dist/builder-eval/index.js.map +0 -1
  18. package/dist/campaign/index.d.ts +0 -7454
  19. package/dist/campaign/index.js +0 -272
  20. package/dist/campaign/index.js.map +0 -1
  21. package/dist/chunk-32BZXMSO.js +0 -3878
  22. package/dist/chunk-32BZXMSO.js.map +0 -1
  23. package/dist/chunk-3A246TSA.js +0 -998
  24. package/dist/chunk-3A246TSA.js.map +0 -1
  25. package/dist/chunk-3RF76KTD.js +0 -84
  26. package/dist/chunk-3RF76KTD.js.map +0 -1
  27. package/dist/chunk-3YYRZDON.js +0 -45
  28. package/dist/chunk-3YYRZDON.js.map +0 -1
  29. package/dist/chunk-4I2E3LLO.js +0 -1030
  30. package/dist/chunk-4I2E3LLO.js.map +0 -1
  31. package/dist/chunk-ARU2PZFM.js +0 -312
  32. package/dist/chunk-ARU2PZFM.js.map +0 -1
  33. package/dist/chunk-BOD4O7OF.js +0 -40
  34. package/dist/chunk-BOD4O7OF.js.map +0 -1
  35. package/dist/chunk-DPZAEKA6.js +0 -880
  36. package/dist/chunk-DPZAEKA6.js.map +0 -1
  37. package/dist/chunk-DTJ6QUQB.js +0 -131
  38. package/dist/chunk-DTJ6QUQB.js.map +0 -1
  39. package/dist/chunk-GGE4NNQT.js +0 -65
  40. package/dist/chunk-GGE4NNQT.js.map +0 -1
  41. package/dist/chunk-H5UD2323.js +0 -286
  42. package/dist/chunk-H5UD2323.js.map +0 -1
  43. package/dist/chunk-HHWE3POT.js +0 -94
  44. package/dist/chunk-HHWE3POT.js.map +0 -1
  45. package/dist/chunk-HKUCJ437.js +0 -787
  46. package/dist/chunk-HKUCJ437.js.map +0 -1
  47. package/dist/chunk-JHCHEVET.js +0 -274
  48. package/dist/chunk-JHCHEVET.js.map +0 -1
  49. package/dist/chunk-JHOJHHU7.js +0 -867
  50. package/dist/chunk-JHOJHHU7.js.map +0 -1
  51. package/dist/chunk-JM2SKQMS.js +0 -750
  52. package/dist/chunk-JM2SKQMS.js.map +0 -1
  53. package/dist/chunk-JN2FCO5W.js +0 -7958
  54. package/dist/chunk-JN2FCO5W.js.map +0 -1
  55. package/dist/chunk-K4DBDHLK.js +0 -158
  56. package/dist/chunk-K4DBDHLK.js.map +0 -1
  57. package/dist/chunk-K6N6XJJX.js +0 -306
  58. package/dist/chunk-K6N6XJJX.js.map +0 -1
  59. package/dist/chunk-MA6HLL3S.js +0 -65
  60. package/dist/chunk-MA6HLL3S.js.map +0 -1
  61. package/dist/chunk-MAZ26DC7.js +0 -99
  62. package/dist/chunk-MAZ26DC7.js.map +0 -1
  63. package/dist/chunk-MOXWMGPC.js +0 -577
  64. package/dist/chunk-MOXWMGPC.js.map +0 -1
  65. package/dist/chunk-NJC7U437.js +0 -626
  66. package/dist/chunk-NJC7U437.js.map +0 -1
  67. package/dist/chunk-NPCTHQIO.js +0 -91
  68. package/dist/chunk-NPCTHQIO.js.map +0 -1
  69. package/dist/chunk-ONWEPEDO.js +0 -57
  70. package/dist/chunk-ONWEPEDO.js.map +0 -1
  71. package/dist/chunk-OYZAPX5G.js +0 -1526
  72. package/dist/chunk-OYZAPX5G.js.map +0 -1
  73. package/dist/chunk-PC4UYEBM.js +0 -166
  74. package/dist/chunk-PC4UYEBM.js.map +0 -1
  75. package/dist/chunk-PICTDURQ.js +0 -766
  76. package/dist/chunk-PICTDURQ.js.map +0 -1
  77. package/dist/chunk-PJQFMIOX.js +0 -1182
  78. package/dist/chunk-PJQFMIOX.js.map +0 -1
  79. package/dist/chunk-PXD6ZFNY.js +0 -1107
  80. package/dist/chunk-PXD6ZFNY.js.map +0 -1
  81. package/dist/chunk-PXE2VKMX.js +0 -140
  82. package/dist/chunk-PXE2VKMX.js.map +0 -1
  83. package/dist/chunk-PZ5AY32C.js +0 -10
  84. package/dist/chunk-PZ5AY32C.js.map +0 -1
  85. package/dist/chunk-QBRSJK47.js +0 -622
  86. package/dist/chunk-QBRSJK47.js.map +0 -1
  87. package/dist/chunk-QWMPPZ3X.js +0 -550
  88. package/dist/chunk-QWMPPZ3X.js.map +0 -1
  89. package/dist/chunk-S3UZOQ5Y.js +0 -328
  90. package/dist/chunk-S3UZOQ5Y.js.map +0 -1
  91. package/dist/chunk-S5TT5R3L.js +0 -2668
  92. package/dist/chunk-S5TT5R3L.js.map +0 -1
  93. package/dist/chunk-T4SQEITX.js +0 -95
  94. package/dist/chunk-T4SQEITX.js.map +0 -1
  95. package/dist/chunk-TT4KNT67.js +0 -124
  96. package/dist/chunk-TT4KNT67.js.map +0 -1
  97. package/dist/chunk-U5CHZ5M3.js +0 -357
  98. package/dist/chunk-U5CHZ5M3.js.map +0 -1
  99. package/dist/chunk-ULOKLHIQ.js +0 -1937
  100. package/dist/chunk-ULOKLHIQ.js.map +0 -1
  101. package/dist/chunk-VI2UW6B6.js +0 -162
  102. package/dist/chunk-VI2UW6B6.js.map +0 -1
  103. package/dist/chunk-VQMK5FMP.js +0 -247
  104. package/dist/chunk-VQMK5FMP.js.map +0 -1
  105. package/dist/chunk-VSMTAMNK.js +0 -53
  106. package/dist/chunk-VSMTAMNK.js.map +0 -1
  107. package/dist/chunk-VZSRQ272.js +0 -149
  108. package/dist/chunk-VZSRQ272.js.map +0 -1
  109. package/dist/chunk-WW2A73HW.js +0 -159
  110. package/dist/chunk-WW2A73HW.js.map +0 -1
  111. package/dist/chunk-X4UCIOTZ.js +0 -136
  112. package/dist/chunk-X4UCIOTZ.js.map +0 -1
  113. package/dist/chunk-XDIRG3TO.js +0 -1266
  114. package/dist/chunk-XDIRG3TO.js.map +0 -1
  115. package/dist/chunk-XJYR7XFV.js +0 -317
  116. package/dist/chunk-XJYR7XFV.js.map +0 -1
  117. package/dist/chunk-ZET2UAYW.js +0 -89
  118. package/dist/chunk-ZET2UAYW.js.map +0 -1
  119. package/dist/chunk-ZZUXHH3R.js +0 -99
  120. package/dist/chunk-ZZUXHH3R.js.map +0 -1
  121. package/dist/cli.d.ts +0 -1
  122. package/dist/cli.js +0 -112
  123. package/dist/cli.js.map +0 -1
  124. package/dist/contract/index.d.ts +0 -4969
  125. package/dist/contract/index.js +0 -1653
  126. package/dist/contract/index.js.map +0 -1
  127. package/dist/control.d.ts +0 -1013
  128. package/dist/control.js +0 -34
  129. package/dist/control.js.map +0 -1
  130. package/dist/fuzz.d.ts +0 -759
  131. package/dist/fuzz.js +0 -714
  132. package/dist/fuzz.js.map +0 -1
  133. package/dist/hosted/index.d.ts +0 -730
  134. package/dist/hosted/index.js +0 -14
  135. package/dist/hosted/index.js.map +0 -1
  136. package/dist/index.d.ts +0 -16780
  137. package/dist/index.js +0 -12168
  138. package/dist/index.js.map +0 -1
  139. package/dist/matrix/index.d.ts +0 -155
  140. package/dist/matrix/index.js +0 -8
  141. package/dist/matrix/index.js.map +0 -1
  142. package/dist/meta-eval/index.d.ts +0 -1030
  143. package/dist/meta-eval/index.js +0 -417
  144. package/dist/meta-eval/index.js.map +0 -1
  145. package/dist/multishot/index.d.ts +0 -579
  146. package/dist/multishot/index.js +0 -589
  147. package/dist/multishot/index.js.map +0 -1
  148. package/dist/openapi.json +0 -992
  149. package/dist/pipelines/index.d.ts +0 -567
  150. package/dist/pipelines/index.js +0 -515
  151. package/dist/pipelines/index.js.map +0 -1
  152. package/dist/reporting.d.ts +0 -1277
  153. package/dist/reporting.js +0 -48
  154. package/dist/reporting.js.map +0 -1
  155. package/dist/rl.d.ts +0 -4092
  156. package/dist/rl.js +0 -1724
  157. package/dist/rl.js.map +0 -1
  158. package/dist/run-campaign-HNFPJET4.js +0 -14
  159. package/dist/run-campaign-HNFPJET4.js.map +0 -1
  160. package/dist/storyboard/index.d.ts +0 -279
  161. package/dist/storyboard/index.js +0 -767
  162. package/dist/storyboard/index.js.map +0 -1
  163. package/dist/trace-attributes.d.ts +0 -52
  164. package/dist/trace-attributes.js +0 -62
  165. package/dist/trace-attributes.js.map +0 -1
  166. package/dist/traces.d.ts +0 -2343
  167. package/dist/traces.js +0 -249
  168. package/dist/traces.js.map +0 -1
  169. package/dist/wire/index.d.ts +0 -1252
  170. package/dist/wire/index.js +0 -81
  171. package/dist/wire/index.js.map +0 -1
@@ -1,155 +0,0 @@
1
- /**
2
- * Validator-output verdict — substrate primitive for "did this output pass,
3
- * and how well?"
4
- *
5
- * Used by:
6
- * - `@tangle-network/agent-eval/matrix` — verdict per cell in the cartesian.
7
- * - `@tangle-network/agent-runtime` — Validator<Output, Verdict = DefaultVerdict>.
8
- * Runtime keeps `Validator` because it's coupled to runtime-shaped
9
- * `ValidationCtx` (iteration, signal, traceEmitter); the verdict TYPE
10
- * itself is a substrate concept and lives here.
11
- *
12
- * Repo layering: agent-eval is the substrate (no upward deps). Both
13
- * agent-runtime and agent-knowledge consume this type FROM agent-eval —
14
- * never the other way around. See CLAUDE.md "Repo layering" for the rule.
15
- */
16
- /**
17
- * Minimal verdict shape — `valid` + `score` are required; `scores` +
18
- * `notes` are optional surface. Validators that need richer shapes
19
- * parameterise `Validator<Output, MyVerdict>` with their own type.
20
- *
21
- * Need structured extras? Extend DefaultVerdict with typed fields — never
22
- * serialize extras into `notes`.
23
- */
24
- interface DefaultVerdict {
25
- /** Whether the output meets the validator's pass criteria. */
26
- valid: boolean;
27
- /** Aggregate score in [0, 1]. Drivers use this for winner selection. */
28
- score: number;
29
- /** Per-dimension scores. Free-form; weighted into `score` by the validator. */
30
- scores?: Record<string, number>;
31
- /** Human-readable rationale; surfaces in trace + final-result `winner.verdict`. */
32
- notes?: string;
33
- }
34
-
35
- /**
36
- * N-axis cartesian matrix over substrate types — types module.
37
- *
38
- * The matrix is a runner + aggregator. It iterates the cartesian product of
39
- * caller-provided axes (any value type — `AgentProfile` from agent-interface,
40
- * `Driver` / `Validator` from agent-runtime, rubric records, thinking levels, anything)
41
- * and aggregates per-axis pass/score/cost summaries. Substrate types are
42
- * imported at the boundary by JSDoc only; the matrix never wraps them.
43
- */
44
-
45
- /** One axis = one dimension to iterate. `V` is the value type — pass any
46
- * substrate type (AgentProfile, Driver, Validator, rubric record). */
47
- interface MatrixAxis<V> {
48
- /** Axis name. Becomes the key in `MatrixResult.byAxis`. */
49
- name: string;
50
- /** Stable id per value. Used as the bucket key in aggregation. */
51
- values: Array<{
52
- id: string;
53
- value: V;
54
- }>;
55
- /** Optional bucket label override. Receives the same `(value, id)` the
56
- * runner stored on the cell; default label is `id`. */
57
- label?: (value: V, id: string) => string;
58
- }
59
- /** A cell carries one picked value from each axis, keyed by axis name. */
60
- interface MatrixCell {
61
- axes: Record<string, {
62
- id: string;
63
- value: unknown;
64
- }>;
65
- /** 0-based replicate index within the same axis combination. */
66
- rep: number;
67
- /** Stable sort key — preserves cartesian order across concurrent execution. */
68
- ordinal: number;
69
- }
70
- interface CellResult<Output> {
71
- output: Output;
72
- verdict: DefaultVerdict;
73
- costUsd: number;
74
- durationMs: number;
75
- runId?: string;
76
- /** Populated when `runCell` threw. The cell contributes 0 to passRate AND
77
- * meanScore regardless of `verdict`. */
78
- error?: {
79
- message: string;
80
- kind: string;
81
- };
82
- }
83
- interface AxisSummary {
84
- axisName: string;
85
- axisValue: string;
86
- cells: number;
87
- passRate: number;
88
- meanScore: number;
89
- p50Score: number;
90
- p90Score: number;
91
- totalCostUsd: number;
92
- meanDurationMs: number;
93
- }
94
- interface MatrixResult<Output> {
95
- cells: Array<{
96
- cell: MatrixCell;
97
- runs: CellResult<Output>[];
98
- }>;
99
- /** `byAxis[axisName][axisValueId] = summary`. Populated only for axes
100
- * named in `aggregateBy` (default = every axis in `axes`). */
101
- byAxis: Record<string, Record<string, AxisSummary>>;
102
- summary: {
103
- totalCells: number;
104
- runsExecuted: number;
105
- /** Cells removed by `filter` plus cells unscheduled after the cost
106
- * ceiling or abort signal tripped. */
107
- cellsSkipped: number;
108
- overallPassRate: number;
109
- overallMeanScore: number;
110
- totalCostUsd: number;
111
- durationMs: number;
112
- };
113
- /** Stable id-like string generated at the end of the run. */
114
- matrixId: string;
115
- }
116
- interface RunAgentMatrixOptions<Output> {
117
- axes: MatrixAxis<unknown>[];
118
- /** User-supplied cell executor. May throw; the matrix captures throws as
119
- * `CellResult.error` and continues. */
120
- runCell: (cell: MatrixCell) => Promise<CellResult<Output>>;
121
- /** Replicates per cell. Default 1. */
122
- reps?: number;
123
- /** Prune cells from the cartesian BEFORE rep expansion. */
124
- filter?: (cell: Omit<MatrixCell, 'rep' | 'ordinal'>) => boolean;
125
- /** Axes to aggregate into `byAxis`. Default: every axis in `axes`. */
126
- aggregateBy?: string[];
127
- /** Max concurrent in-flight `runCell` invocations. Default 4. */
128
- maxConcurrency?: number;
129
- /** Cumulative-cost abort threshold (USD). When the running sum of
130
- * `result.costUsd` crosses this value, no new cells are scheduled.
131
- * In-flight cells finish. Default `Infinity`. */
132
- costCeiling?: number;
133
- /** Fires once per executed cell, after its promise settles. */
134
- onCellComplete?: (cell: MatrixCell, result: CellResult<Output>) => void;
135
- /** External cancellation. Aborts in-flight cells via a forwarded signal
136
- * and suppresses scheduling of new ones. */
137
- signal?: AbortSignal;
138
- }
139
-
140
- /**
141
- * N-axis cartesian runner.
142
- *
143
- * Expansion order: cartesian over `axes` in declared order, then `reps` as the
144
- * inner-most dim → `ordinal = (cartIdx * reps) + rep`. The returned
145
- * `cells[]` is sorted by `ordinal` so concurrent execution does not reorder
146
- * the output.
147
- *
148
- * Scheduling is a sliding window of in-flight promises capped at
149
- * `maxConcurrency`. The window stops admitting new cells when the cost
150
- * ceiling trips or the abort signal fires; in-flight cells finish.
151
- */
152
-
153
- declare function runAgentMatrix<Output>(opts: RunAgentMatrixOptions<Output>): Promise<MatrixResult<Output>>;
154
-
155
- export { type AxisSummary, type CellResult, type DefaultVerdict, type MatrixAxis, type MatrixCell, type MatrixResult, type RunAgentMatrixOptions, runAgentMatrix };
@@ -1,8 +0,0 @@
1
- import {
2
- runAgentMatrix
3
- } from "../chunk-JHCHEVET.js";
4
- import "../chunk-PZ5AY32C.js";
5
- export {
6
- runAgentMatrix
7
- };
8
- //# sourceMappingURL=index.js.map
@@ -1 +0,0 @@
1
- {"version":3,"sources":[],"sourcesContent":[],"mappings":"","names":[]}