@pentatonic-ai/ai-agent-sdk 0.9.5 → 0.10.0

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 (131) hide show
  1. package/README.md +3 -3
  2. package/bin/cli.js +1 -1
  3. package/bin/commands/config.js +1 -1
  4. package/dist/index.cjs +39 -72
  5. package/dist/index.js +36 -69
  6. package/package.json +10 -3
  7. package/packages/doctor/src/checks/local-memory.js +2 -2
  8. package/packages/memory/README.md +2 -2
  9. package/packages/memory/openclaw-plugin/README.md +2 -2
  10. package/packages/memory/openclaw-plugin/openclaw.plugin.json +1 -1
  11. package/packages/memory/package-lock.json +49 -33
  12. package/packages/memory/package.json +4 -1
  13. package/packages/memory/src/__tests__/engine.test.js +40 -5
  14. package/packages/memory/src/engine.js +38 -3
  15. package/packages/memory/src/server.js +2 -2
  16. package/packages/memory-engine-v2/.env.example +30 -0
  17. package/packages/memory-engine-v2/README.md +125 -0
  18. package/packages/memory-engine-v2/compat/Dockerfile +11 -0
  19. package/packages/memory-engine-v2/compat/requirements.txt +6 -0
  20. package/packages/memory-engine-v2/compat/server.py +1047 -0
  21. package/packages/memory-engine-v2/docker-compose.aws.yml +78 -0
  22. package/packages/memory-engine-v2/docker-compose.yml +206 -0
  23. package/packages/memory-engine-v2/extractor-async/Dockerfile +14 -0
  24. package/packages/memory-engine-v2/extractor-async/confidence.py +62 -0
  25. package/packages/memory-engine-v2/extractor-async/noise_filter.py +144 -0
  26. package/packages/memory-engine-v2/extractor-async/requirements.txt +2 -0
  27. package/packages/memory-engine-v2/extractor-async/test_confidence.py +76 -0
  28. package/packages/memory-engine-v2/extractor-async/test_noise_filter.py +177 -0
  29. package/packages/memory-engine-v2/extractor-async/worker.py +797 -0
  30. package/packages/memory-engine-v2/extractor-sync/Dockerfile +11 -0
  31. package/packages/memory-engine-v2/extractor-sync/requirements.txt +4 -0
  32. package/packages/memory-engine-v2/extractor-sync/server.py +424 -0
  33. package/packages/memory-engine-v2/org-model/migrations/001_init.sql +390 -0
  34. package/packages/memory-engine-v2/tests/e2e_smoke.py +356 -0
  35. package/packages/memory-engine-v2/tests/fixtures/generate_synthetic_corpus.py +758 -0
  36. package/packages/memory-engine/.env.example +0 -13
  37. package/packages/memory-engine/MIGRATION.md +0 -219
  38. package/packages/memory-engine/README.md +0 -145
  39. package/packages/memory-engine/bench/README.md +0 -99
  40. package/packages/memory-engine/bench/scorecards-engine/agent-coding__pentatonic-baseline__20260427-142523.json +0 -1115
  41. package/packages/memory-engine/bench/scorecards-engine/chat-recall__pentatonic-baseline__20260427-142648.json +0 -819
  42. package/packages/memory-engine/bench/scorecards-engine/circular-economy__pentatonic-baseline__20260427-142757.json +0 -1278
  43. package/packages/memory-engine/bench/scorecards-engine/customer-support__pentatonic-baseline__20260427-142900.json +0 -1018
  44. package/packages/memory-engine/bench/scorecards-engine/marketplace-ops__pentatonic-baseline__20260427-142957.json +0 -1038
  45. package/packages/memory-engine/bench/scorecards-engine/product-catalogue__pentatonic-baseline__20260427-143122.json +0 -961
  46. package/packages/memory-engine/bench/scorecards-engine-via-docker/agent-coding__pentatonic-memory__20260427-161812.json +0 -1115
  47. package/packages/memory-engine/bench/scorecards-engine-via-docker/chat-recall__pentatonic-memory__20260427-161701.json +0 -819
  48. package/packages/memory-engine/bench/scorecards-engine-via-docker/circular-economy__pentatonic-memory__20260427-161713.json +0 -1278
  49. package/packages/memory-engine/bench/scorecards-engine-via-docker/customer-support__pentatonic-memory__20260427-161723.json +0 -1018
  50. package/packages/memory-engine/bench/scorecards-engine-via-docker/marketplace-ops__pentatonic-memory__20260427-161732.json +0 -1038
  51. package/packages/memory-engine/bench/scorecards-engine-via-docker/product-catalogue__pentatonic-memory__20260427-161741.json +0 -937
  52. package/packages/memory-engine/bench/scorecards-engine-via-l2-7-layer-populated/agent-coding__pentatonic-memory__20260427-184718.json +0 -1115
  53. package/packages/memory-engine/bench/scorecards-engine-via-l2-7-layer-populated/chat-recall__pentatonic-memory__20260427-184614.json +0 -819
  54. package/packages/memory-engine/bench/scorecards-engine-via-l2-7-layer-populated/circular-economy__pentatonic-memory__20260427-184809.json +0 -1278
  55. package/packages/memory-engine/bench/scorecards-engine-via-l2-7-layer-populated/customer-support__pentatonic-memory__20260427-184854.json +0 -1018
  56. package/packages/memory-engine/bench/scorecards-engine-via-l2-7-layer-populated/marketplace-ops__pentatonic-memory__20260427-184929.json +0 -1038
  57. package/packages/memory-engine/bench/scorecards-engine-via-l2-7-layer-populated/product-catalogue__pentatonic-memory__20260427-185015.json +0 -961
  58. package/packages/memory-engine/bench/scorecards-engine-via-l2-empty-layers/agent-coding__pentatonic-memory__20260427-175252.json +0 -1115
  59. package/packages/memory-engine/bench/scorecards-engine-via-l2-empty-layers/chat-recall__pentatonic-memory__20260427-175312.json +0 -819
  60. package/packages/memory-engine/bench/scorecards-engine-via-l2-empty-layers/circular-economy__pentatonic-memory__20260427-175335.json +0 -1278
  61. package/packages/memory-engine/bench/scorecards-engine-via-l2-empty-layers/customer-support__pentatonic-memory__20260427-175355.json +0 -1018
  62. package/packages/memory-engine/bench/scorecards-engine-via-l2-empty-layers/marketplace-ops__pentatonic-memory__20260427-175413.json +0 -1038
  63. package/packages/memory-engine/bench/scorecards-engine-via-l2-empty-layers/product-catalogue__pentatonic-memory__20260427-175430.json +0 -883
  64. package/packages/memory-engine/bench/scorecards-engine-via-shim/agent-coding__pentatonic-memory__20260427-155409.json +0 -1115
  65. package/packages/memory-engine/bench/scorecards-engine-via-shim/chat-recall__pentatonic-memory__20260427-155421.json +0 -819
  66. package/packages/memory-engine/bench/scorecards-engine-via-shim/circular-economy__pentatonic-memory__20260427-155433.json +0 -1278
  67. package/packages/memory-engine/bench/scorecards-engine-via-shim/customer-support__pentatonic-memory__20260427-155443.json +0 -1018
  68. package/packages/memory-engine/bench/scorecards-engine-via-shim/marketplace-ops__pentatonic-memory__20260427-155453.json +0 -1038
  69. package/packages/memory-engine/bench/scorecards-engine-via-shim/product-catalogue__pentatonic-memory__20260427-155503.json +0 -937
  70. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/agent-coding__pentatonic-memory-latest__20260427-145103.json +0 -1115
  71. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/agent-coding__pentatonic-memory__20260427-144909.json +0 -1115
  72. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/chat-recall__pentatonic-memory-latest__20260427-145153.json +0 -819
  73. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/chat-recall__pentatonic-memory__20260427-145120.json +0 -542
  74. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/circular-economy__pentatonic-memory-latest__20260427-145313.json +0 -1278
  75. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/circular-economy__pentatonic-memory__20260427-145207.json +0 -894
  76. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/customer-support__pentatonic-memory-latest__20260427-145412.json +0 -1018
  77. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/customer-support__pentatonic-memory__20260427-145327.json +0 -680
  78. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/marketplace-ops__pentatonic-memory-latest__20260427-145517.json +0 -1038
  79. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/marketplace-ops__pentatonic-memory__20260427-145422.json +0 -693
  80. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/product-catalogue__pentatonic-memory-latest__20260427-145616.json +0 -961
  81. package/packages/memory-engine/bench/scorecards-pentatonic-baseline/product-catalogue__pentatonic-memory__20260427-145528.json +0 -727
  82. package/packages/memory-engine/compat/Dockerfile +0 -22
  83. package/packages/memory-engine/compat/server.py +0 -1255
  84. package/packages/memory-engine/docker-compose.test.yml +0 -59
  85. package/packages/memory-engine/docker-compose.yml +0 -240
  86. package/packages/memory-engine/engine/README.md +0 -52
  87. package/packages/memory-engine/engine/l2-hybridrag-proxy.py +0 -1543
  88. package/packages/memory-engine/engine/l5-comms-layer.py +0 -663
  89. package/packages/memory-engine/engine/l6-document-store.py +0 -1018
  90. package/packages/memory-engine/engine/services/_shared/__init__.py +0 -1
  91. package/packages/memory-engine/engine/services/_shared/embed_provider.py +0 -468
  92. package/packages/memory-engine/engine/services/l2/Dockerfile +0 -50
  93. package/packages/memory-engine/engine/services/l2/init_databases.py +0 -81
  94. package/packages/memory-engine/engine/services/l2/l2-hybridrag-proxy.py +0 -2721
  95. package/packages/memory-engine/engine/services/l5/Dockerfile +0 -11
  96. package/packages/memory-engine/engine/services/l5/l5-comms-layer.py +0 -808
  97. package/packages/memory-engine/engine/services/l6/Dockerfile +0 -30
  98. package/packages/memory-engine/engine/services/l6/l6-document-store.py +0 -1221
  99. package/packages/memory-engine/engine/services/nv-embed/Dockerfile +0 -28
  100. package/packages/memory-engine/engine/services/nv-embed/server.py +0 -152
  101. package/packages/memory-engine/pme_memory/__init__.py +0 -0
  102. package/packages/memory-engine/pme_memory/__main__.py +0 -129
  103. package/packages/memory-engine/pme_memory/artifacts.py +0 -95
  104. package/packages/memory-engine/pme_memory/embed.py +0 -74
  105. package/packages/memory-engine/pme_memory/health.py +0 -36
  106. package/packages/memory-engine/pme_memory/hygiene.py +0 -159
  107. package/packages/memory-engine/pme_memory/indexer.py +0 -200
  108. package/packages/memory-engine/pme_memory/needs.py +0 -55
  109. package/packages/memory-engine/pme_memory/provenance.py +0 -80
  110. package/packages/memory-engine/pme_memory/scoring.py +0 -168
  111. package/packages/memory-engine/pme_memory/search.py +0 -52
  112. package/packages/memory-engine/pme_memory/store.py +0 -86
  113. package/packages/memory-engine/pme_memory/synthesis.py +0 -114
  114. package/packages/memory-engine/pyproject.toml +0 -65
  115. package/packages/memory-engine/scripts/kg-extractor.py +0 -557
  116. package/packages/memory-engine/scripts/kg-preflexor-v2.py +0 -738
  117. package/packages/memory-engine/scripts/wipe-legacy-l3-entities.py +0 -128
  118. package/packages/memory-engine/tests/e2e_arena.sh +0 -259
  119. package/packages/memory-engine/tests/embed_stub/Dockerfile +0 -13
  120. package/packages/memory-engine/tests/embed_stub/server.py +0 -80
  121. package/packages/memory-engine/tests/test_aggregate.py +0 -333
  122. package/packages/memory-engine/tests/test_api_contract.sh +0 -57
  123. package/packages/memory-engine/tests/test_arena_safety.py +0 -232
  124. package/packages/memory-engine/tests/test_channel_stat_reader.py +0 -437
  125. package/packages/memory-engine/tests/test_channel_stat_rollups.py +0 -308
  126. package/packages/memory-engine/tests/test_compat_nv_embed_probe.py +0 -48
  127. package/packages/memory-engine/tests/test_embed_provider.py +0 -492
  128. package/packages/memory-engine/tests/test_l2_qmd_vec_search.py +0 -280
  129. package/packages/memory-engine/tests/test_l3_arena_isolation.py +0 -412
  130. package/packages/memory-engine/tests/test_l6_module_load.py +0 -84
  131. package/packages/memory-engine/tests/test_people_list_reader.py +0 -432
@@ -1,819 +0,0 @@
1
- {
2
- "bench": "chat-recall",
3
- "stack": "pentatonic-memory-latest",
4
- "n_tasks": 16,
5
- "n_correct": 0,
6
- "accuracy": 0.0,
7
- "mean_score": 0.0,
8
- "p50_search_ms": 35.576166497776285,
9
- "p95_search_ms": 37.78688400052488,
10
- "total_tokens_in": 0,
11
- "total_tokens_out": 0,
12
- "total_usd": 0.0,
13
- "by_tag": {
14
- "factoid": {
15
- "n": 6,
16
- "mean_score": 0.0,
17
- "accuracy": 0.0
18
- },
19
- "owner": {
20
- "n": 4,
21
- "mean_score": 0.0,
22
- "accuracy": 0.0
23
- },
24
- "temporal": {
25
- "n": 4,
26
- "mean_score": 0.0,
27
- "accuracy": 0.0
28
- },
29
- "contradiction": {
30
- "n": 2,
31
- "mean_score": 0.0,
32
- "accuracy": 0.0
33
- },
34
- "status": {
35
- "n": 4,
36
- "mean_score": 0.0,
37
- "accuracy": 0.0
38
- },
39
- "multi-hop": {
40
- "n": 4,
41
- "mean_score": 0.0,
42
- "accuracy": 0.0
43
- }
44
- },
45
- "extra": {
46
- "ingest_ms": 32138.345434999792,
47
- "grading": "substring",
48
- "limit": 3,
49
- "tokens": {
50
- "corpus_tokens": 513,
51
- "query_tokens": 140,
52
- "context_tokens": 630,
53
- "retrieval_tokens": 770,
54
- "naive_tokens": 8348,
55
- "saved_tokens": 7578,
56
- "reduction_pct": 0.907762338284619,
57
- "mean_retrieval_tokens_per_task": 48.125,
58
- "tokenizer": "cl100k_base",
59
- "per_task": {
60
- "who-owns-atlas": {
61
- "query": 5,
62
- "context": 34,
63
- "retrieval": 39,
64
- "judge_in": 0,
65
- "judge_out": 0,
66
- "judge_latency_ms": 0.0
67
- },
68
- "who-owns-borealis": {
69
- "query": 7,
70
- "context": 34,
71
- "retrieval": 41,
72
- "judge_in": 0,
73
- "judge_out": 0,
74
- "judge_latency_ms": 0.0
75
- },
76
- "who-owns-cirrus": {
77
- "query": 6,
78
- "context": 49,
79
- "retrieval": 55,
80
- "judge_in": 0,
81
- "judge_out": 0,
82
- "judge_latency_ms": 0.0
83
- },
84
- "who-owns-dune": {
85
- "query": 6,
86
- "context": 31,
87
- "retrieval": 37,
88
- "judge_in": 0,
89
- "judge_out": 0,
90
- "judge_latency_ms": 0.0
91
- },
92
- "current-deadline-atlas": {
93
- "query": 8,
94
- "context": 35,
95
- "retrieval": 43,
96
- "judge_in": 0,
97
- "judge_out": 0,
98
- "judge_latency_ms": 0.0
99
- },
100
- "current-deadline-borealis": {
101
- "query": 10,
102
- "context": 34,
103
- "retrieval": 44,
104
- "judge_in": 0,
105
- "judge_out": 0,
106
- "judge_latency_ms": 0.0
107
- },
108
- "current-deadline-cirrus": {
109
- "query": 9,
110
- "context": 42,
111
- "retrieval": 51,
112
- "judge_in": 0,
113
- "judge_out": 0,
114
- "judge_latency_ms": 0.0
115
- },
116
- "current-deadline-dune": {
117
- "query": 9,
118
- "context": 31,
119
- "retrieval": 40,
120
- "judge_in": 0,
121
- "judge_out": 0,
122
- "judge_latency_ms": 0.0
123
- },
124
- "status-atlas": {
125
- "query": 8,
126
- "context": 33,
127
- "retrieval": 41,
128
- "judge_in": 0,
129
- "judge_out": 0,
130
- "judge_latency_ms": 0.0
131
- },
132
- "status-borealis": {
133
- "query": 10,
134
- "context": 42,
135
- "retrieval": 52,
136
- "judge_in": 0,
137
- "judge_out": 0,
138
- "judge_latency_ms": 0.0
139
- },
140
- "status-cirrus": {
141
- "query": 9,
142
- "context": 50,
143
- "retrieval": 59,
144
- "judge_in": 0,
145
- "judge_out": 0,
146
- "judge_latency_ms": 0.0
147
- },
148
- "status-dune": {
149
- "query": 9,
150
- "context": 35,
151
- "retrieval": 44,
152
- "judge_in": 0,
153
- "judge_out": 0,
154
- "judge_latency_ms": 0.0
155
- },
156
- "multihop-atlas": {
157
- "query": 10,
158
- "context": 42,
159
- "retrieval": 52,
160
- "judge_in": 0,
161
- "judge_out": 0,
162
- "judge_latency_ms": 0.0
163
- },
164
- "multihop-borealis": {
165
- "query": 12,
166
- "context": 50,
167
- "retrieval": 62,
168
- "judge_in": 0,
169
- "judge_out": 0,
170
- "judge_latency_ms": 0.0
171
- },
172
- "multihop-cirrus": {
173
- "query": 11,
174
- "context": 50,
175
- "retrieval": 61,
176
- "judge_in": 0,
177
- "judge_out": 0,
178
- "judge_latency_ms": 0.0
179
- },
180
- "multihop-dune": {
181
- "query": 11,
182
- "context": 38,
183
- "retrieval": 49,
184
- "judge_in": 0,
185
- "judge_out": 0,
186
- "judge_latency_ms": 0.0
187
- }
188
- },
189
- "judge_tokens_in": 0,
190
- "judge_tokens_out": 0,
191
- "judge_calls": 0,
192
- "judge_mean_latency_ms": 0.0
193
- },
194
- "cost_usd": {
195
- "assumed_completion_tokens_per_task": 100,
196
- "rates": {
197
- "input_per_1k": 0.0025,
198
- "output_per_1k": 0.01,
199
- "model": "gpt-4o"
200
- },
201
- "retrieval_usd_in": 0.001925,
202
- "retrieval_usd_out": 0.016,
203
- "retrieval_usd_total": 0.017925,
204
- "naive_usd_total": 0.03687,
205
- "saved_usd": 0.018945,
206
- "saved_usd_per_1k_tasks": 1.1840625
207
- }
208
- },
209
- "task_results": [
210
- {
211
- "task_id": "who-owns-atlas",
212
- "query": "Who owns project Atlas?",
213
- "answer": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nbilling-svc API keys \u2014 jordan\n---\nsearch-svc API keys \u2014 alex",
214
- "hits": [
215
- {
216
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
217
- "score": 0.7094495479744956,
218
- "source": "pentatonic-memory",
219
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
220
- },
221
- {
222
- "text": "billing-svc API keys \u2014 jordan",
223
- "score": 0.705149493119418,
224
- "source": "pentatonic-memory",
225
- "doc_id": "mem_80182b2e-cbba-4642-b6fd-b8d3fbd6e617"
226
- },
227
- {
228
- "text": "search-svc API keys \u2014 alex",
229
- "score": 0.6992094186827446,
230
- "source": "pentatonic-memory",
231
- "doc_id": "mem_730b4496-40b1-499f-952a-739ae3912826"
232
- }
233
- ],
234
- "correct": false,
235
- "score": 0.0,
236
- "grading_notes": "missing 1/1: ['Alice']",
237
- "search_time_ms": 29.684740991797298,
238
- "generation_time_ms": 0.0,
239
- "tokens_in": 0,
240
- "tokens_out": 0,
241
- "retrieval_tokens": 39,
242
- "query_tokens": 5,
243
- "context_tokens": 34,
244
- "judge_tokens_in": 0,
245
- "judge_tokens_out": 0,
246
- "judge_latency_ms": 0.0
247
- },
248
- {
249
- "task_id": "who-owns-borealis",
250
- "query": "Who owns project Borealis?",
251
- "answer": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nsearch-svc API keys \u2014 alex\n---\nbilling-svc API keys \u2014 jordan",
252
- "hits": [
253
- {
254
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
255
- "score": 0.6940128053794369,
256
- "source": "pentatonic-memory",
257
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
258
- },
259
- {
260
- "text": "search-svc API keys \u2014 alex",
261
- "score": 0.6928977121880233,
262
- "source": "pentatonic-memory",
263
- "doc_id": "mem_730b4496-40b1-499f-952a-739ae3912826"
264
- },
265
- {
266
- "text": "billing-svc API keys \u2014 jordan",
267
- "score": 0.6927844693031652,
268
- "source": "pentatonic-memory",
269
- "doc_id": "mem_80182b2e-cbba-4642-b6fd-b8d3fbd6e617"
270
- }
271
- ],
272
- "correct": false,
273
- "score": 0.0,
274
- "grading_notes": "missing 1/1: ['Clara']",
275
- "search_time_ms": 27.342910005245358,
276
- "generation_time_ms": 0.0,
277
- "tokens_in": 0,
278
- "tokens_out": 0,
279
- "retrieval_tokens": 41,
280
- "query_tokens": 7,
281
- "context_tokens": 34,
282
- "judge_tokens_in": 0,
283
- "judge_tokens_out": 0,
284
- "judge_latency_ms": 0.0
285
- },
286
- {
287
- "task_id": "who-owns-cirrus",
288
- "query": "Who owns project Cirrus?",
289
- "answer": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\norder-svc marks order as paid, emits OrderPaid on pubsub\n---\nhandle_webhook verifies signature, enqueues internal InvoicePaid event on order-svc",
290
- "hits": [
291
- {
292
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
293
- "score": 0.7071406152219076,
294
- "source": "pentatonic-memory",
295
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
296
- },
297
- {
298
- "text": "order-svc marks order as paid, emits OrderPaid on pubsub",
299
- "score": 0.6962335336552081,
300
- "source": "pentatonic-memory",
301
- "doc_id": "mem_f1083a75-3da4-4a35-bce4-db4a29b46034"
302
- },
303
- {
304
- "text": "handle_webhook verifies signature, enqueues internal InvoicePaid event on order-svc",
305
- "score": 0.6938935982281584,
306
- "source": "pentatonic-memory",
307
- "doc_id": "mem_b6596977-9a91-4469-a8c5-3872edbbbf1c"
308
- }
309
- ],
310
- "correct": false,
311
- "score": 0.0,
312
- "grading_notes": "missing 1/1: ['Diego']",
313
- "search_time_ms": 23.93999500782229,
314
- "generation_time_ms": 0.0,
315
- "tokens_in": 0,
316
- "tokens_out": 0,
317
- "retrieval_tokens": 55,
318
- "query_tokens": 6,
319
- "context_tokens": 49,
320
- "judge_tokens_in": 0,
321
- "judge_tokens_out": 0,
322
- "judge_latency_ms": 0.0
323
- },
324
- {
325
- "task_id": "who-owns-dune",
326
- "query": "Who owns project Dune?",
327
- "answer": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nbilling.new_invoice_pdf\n---\nbilling-svc API keys \u2014 jordan",
328
- "hits": [
329
- {
330
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
331
- "score": 0.6957184482713455,
332
- "source": "pentatonic-memory",
333
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
334
- },
335
- {
336
- "text": "billing.new_invoice_pdf",
337
- "score": 0.6797076893489076,
338
- "source": "pentatonic-memory",
339
- "doc_id": "mem_ca8bdac3-12ca-4bae-b5a2-fb9c2d069173"
340
- },
341
- {
342
- "text": "billing-svc API keys \u2014 jordan",
343
- "score": 0.6776694304617411,
344
- "source": "pentatonic-memory",
345
- "doc_id": "mem_80182b2e-cbba-4642-b6fd-b8d3fbd6e617"
346
- }
347
- ],
348
- "correct": false,
349
- "score": 0.0,
350
- "grading_notes": "missing 1/1: ['Farid']",
351
- "search_time_ms": 25.87755100103095,
352
- "generation_time_ms": 0.0,
353
- "tokens_in": 0,
354
- "tokens_out": 0,
355
- "retrieval_tokens": 37,
356
- "query_tokens": 6,
357
- "context_tokens": 31,
358
- "judge_tokens_in": 0,
359
- "judge_tokens_out": 0,
360
- "judge_latency_ms": 0.0
361
- },
362
- {
363
- "task_id": "current-deadline-atlas",
364
- "query": "What is the current deadline for Atlas?",
365
- "answer": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nbilling-svc API keys \u2014 jordan\n---\nkubectl rollout restart deployment/billing-svc",
366
- "hits": [
367
- {
368
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
369
- "score": 0.7275709883182019,
370
- "source": "pentatonic-memory",
371
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
372
- },
373
- {
374
- "text": "billing-svc API keys \u2014 jordan",
375
- "score": 0.7167343487691263,
376
- "source": "pentatonic-memory",
377
- "doc_id": "mem_80182b2e-cbba-4642-b6fd-b8d3fbd6e617"
378
- },
379
- {
380
- "text": "kubectl rollout restart deployment/billing-svc",
381
- "score": 0.7163043601685071,
382
- "source": "pentatonic-memory",
383
- "doc_id": "mem_049e0812-46ed-4db7-84d8-c4a121a1d19c"
384
- }
385
- ],
386
- "correct": false,
387
- "score": 0.0,
388
- "grading_notes": "missing 1/1: ['2026-03-17']",
389
- "search_time_ms": 37.78688400052488,
390
- "generation_time_ms": 0.0,
391
- "tokens_in": 0,
392
- "tokens_out": 0,
393
- "retrieval_tokens": 43,
394
- "query_tokens": 8,
395
- "context_tokens": 35,
396
- "judge_tokens_in": 0,
397
- "judge_tokens_out": 0,
398
- "judge_latency_ms": 0.0
399
- },
400
- {
401
- "task_id": "current-deadline-borealis",
402
- "query": "What is the current deadline for Borealis?",
403
- "answer": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nbilling-svc API keys \u2014 jordan\n---\nsearch-svc API keys \u2014 alex",
404
- "hits": [
405
- {
406
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
407
- "score": 0.7083399225416024,
408
- "source": "pentatonic-memory",
409
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
410
- },
411
- {
412
- "text": "billing-svc API keys \u2014 jordan",
413
- "score": 0.6995845993743947,
414
- "source": "pentatonic-memory",
415
- "doc_id": "mem_80182b2e-cbba-4642-b6fd-b8d3fbd6e617"
416
- },
417
- {
418
- "text": "search-svc API keys \u2014 alex",
419
- "score": 0.6989382834457598,
420
- "source": "pentatonic-memory",
421
- "doc_id": "mem_730b4496-40b1-499f-952a-739ae3912826"
422
- }
423
- ],
424
- "correct": false,
425
- "score": 0.0,
426
- "grading_notes": "missing 1/1: ['2026-02-28']",
427
- "search_time_ms": 37.3624479980208,
428
- "generation_time_ms": 0.0,
429
- "tokens_in": 0,
430
- "tokens_out": 0,
431
- "retrieval_tokens": 44,
432
- "query_tokens": 10,
433
- "context_tokens": 34,
434
- "judge_tokens_in": 0,
435
- "judge_tokens_out": 0,
436
- "judge_latency_ms": 0.0
437
- },
438
- {
439
- "task_id": "current-deadline-cirrus",
440
- "query": "What is the current deadline for Cirrus?",
441
- "answer": "The user's latest Cirrus deadline is April 8, 2026.\n---\nIf secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nkubectl rollout restart deployment/billing-svc",
442
- "hits": [
443
- {
444
- "text": "The user's latest Cirrus deadline is April 8, 2026.",
445
- "score": 0.7558460678895833,
446
- "source": "pentatonic-memory",
447
- "doc_id": "chat-deadline2-cirrus"
448
- },
449
- {
450
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
451
- "score": 0.7260960333461852,
452
- "source": "pentatonic-memory",
453
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
454
- },
455
- {
456
- "text": "kubectl rollout restart deployment/billing-svc",
457
- "score": 0.7114582739003221,
458
- "source": "pentatonic-memory",
459
- "doc_id": "mem_049e0812-46ed-4db7-84d8-c4a121a1d19c"
460
- }
461
- ],
462
- "correct": false,
463
- "score": 0.0,
464
- "grading_notes": "missing 1/1: ['2026-04-08']",
465
- "search_time_ms": 33.95489399554208,
466
- "generation_time_ms": 0.0,
467
- "tokens_in": 0,
468
- "tokens_out": 0,
469
- "retrieval_tokens": 51,
470
- "query_tokens": 9,
471
- "context_tokens": 42,
472
- "judge_tokens_in": 0,
473
- "judge_tokens_out": 0,
474
- "judge_latency_ms": 0.0
475
- },
476
- {
477
- "task_id": "current-deadline-dune",
478
- "query": "What is the current deadline for Dune?",
479
- "answer": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nbilling.new_invoice_pdf\n---\nkubectl rollout restart deployment/billing-svc",
480
- "hits": [
481
- {
482
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
483
- "score": 0.7106671210223783,
484
- "source": "pentatonic-memory",
485
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
486
- },
487
- {
488
- "text": "billing.new_invoice_pdf",
489
- "score": 0.7030500225794961,
490
- "source": "pentatonic-memory",
491
- "doc_id": "mem_ca8bdac3-12ca-4bae-b5a2-fb9c2d069173"
492
- },
493
- {
494
- "text": "kubectl rollout restart deployment/billing-svc",
495
- "score": 0.696777937772653,
496
- "source": "pentatonic-memory",
497
- "doc_id": "mem_049e0812-46ed-4db7-84d8-c4a121a1d19c"
498
- }
499
- ],
500
- "correct": false,
501
- "score": 0.0,
502
- "grading_notes": "missing 1/1: ['2026-05-20']",
503
- "search_time_ms": 35.3459480102174,
504
- "generation_time_ms": 0.0,
505
- "tokens_in": 0,
506
- "tokens_out": 0,
507
- "retrieval_tokens": 40,
508
- "query_tokens": 9,
509
- "context_tokens": 31,
510
- "judge_tokens_in": 0,
511
- "judge_tokens_out": 0,
512
- "judge_latency_ms": 0.0
513
- },
514
- {
515
- "task_id": "status-atlas",
516
- "query": "What's the latest status of Atlas?",
517
- "answer": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nThe Atlas project is currently in development.\n---\nsearch-svc API keys \u2014 alex",
518
- "hits": [
519
- {
520
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
521
- "score": 0.7254432691030256,
522
- "source": "pentatonic-memory",
523
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
524
- },
525
- {
526
- "text": "The Atlas project is currently in development.",
527
- "score": 0.7171795355296468,
528
- "source": "pentatonic-memory",
529
- "doc_id": "chat-deadline2-atlas"
530
- },
531
- {
532
- "text": "search-svc API keys \u2014 alex",
533
- "score": 0.7042820502266752,
534
- "source": "pentatonic-memory",
535
- "doc_id": "mem_730b4496-40b1-499f-952a-739ae3912826"
536
- }
537
- ],
538
- "correct": false,
539
- "score": 0.0,
540
- "grading_notes": "missing 1/1: ['on track']",
541
- "search_time_ms": 36.89389998908155,
542
- "generation_time_ms": 0.0,
543
- "tokens_in": 0,
544
- "tokens_out": 0,
545
- "retrieval_tokens": 41,
546
- "query_tokens": 8,
547
- "context_tokens": 33,
548
- "judge_tokens_in": 0,
549
- "judge_tokens_out": 0,
550
- "judge_latency_ms": 0.0
551
- },
552
- {
553
- "task_id": "status-borealis",
554
- "query": "What's the latest status of Borealis?",
555
- "answer": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nsearch-svc API keys \u2014 alex\n---\nhandle_webhook verifies signature, enqueues internal InvoicePaid event on order-svc",
556
- "hits": [
557
- {
558
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
559
- "score": 0.6994621643226522,
560
- "source": "pentatonic-memory",
561
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
562
- },
563
- {
564
- "text": "search-svc API keys \u2014 alex",
565
- "score": 0.6903014011152464,
566
- "source": "pentatonic-memory",
567
- "doc_id": "mem_730b4496-40b1-499f-952a-739ae3912826"
568
- },
569
- {
570
- "text": "handle_webhook verifies signature, enqueues internal InvoicePaid event on order-svc",
571
- "score": 0.6872532831436886,
572
- "source": "pentatonic-memory",
573
- "doc_id": "mem_b6596977-9a91-4469-a8c5-3872edbbbf1c"
574
- }
575
- ],
576
- "correct": false,
577
- "score": 0.0,
578
- "grading_notes": "missing 1/1: ['blocked on vendor']",
579
- "search_time_ms": 36.008898983709514,
580
- "generation_time_ms": 0.0,
581
- "tokens_in": 0,
582
- "tokens_out": 0,
583
- "retrieval_tokens": 52,
584
- "query_tokens": 10,
585
- "context_tokens": 42,
586
- "judge_tokens_in": 0,
587
- "judge_tokens_out": 0,
588
- "judge_latency_ms": 0.0
589
- },
590
- {
591
- "task_id": "status-cirrus",
592
- "query": "What's the latest status of Cirrus?",
593
- "answer": "The user's latest Cirrus deadline is April 8, 2026.\n---\nIf secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nhandle_webhook verifies signature, enqueues internal InvoicePaid event on order-svc",
594
- "hits": [
595
- {
596
- "text": "The user's latest Cirrus deadline is April 8, 2026.",
597
- "score": 0.7447236868936458,
598
- "source": "pentatonic-memory",
599
- "doc_id": "chat-deadline2-cirrus"
600
- },
601
- {
602
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
603
- "score": 0.7199085720912451,
604
- "source": "pentatonic-memory",
605
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
606
- },
607
- {
608
- "text": "handle_webhook verifies signature, enqueues internal InvoicePaid event on order-svc",
609
- "score": 0.701595044183101,
610
- "source": "pentatonic-memory",
611
- "doc_id": "mem_b6596977-9a91-4469-a8c5-3872edbbbf1c"
612
- }
613
- ],
614
- "correct": false,
615
- "score": 0.0,
616
- "grading_notes": "missing 1/1: ['scoping']",
617
- "search_time_ms": 35.968545998912305,
618
- "generation_time_ms": 0.0,
619
- "tokens_in": 0,
620
- "tokens_out": 0,
621
- "retrieval_tokens": 59,
622
- "query_tokens": 9,
623
- "context_tokens": 50,
624
- "judge_tokens_in": 0,
625
- "judge_tokens_out": 0,
626
- "judge_latency_ms": 0.0
627
- },
628
- {
629
- "task_id": "status-dune",
630
- "query": "What's the latest status of Dune?",
631
- "answer": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nbilling.new_invoice_pdf\n---\nStripe posts a invoice.payment_succeeded webhook to billing-svc",
632
- "hits": [
633
- {
634
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
635
- "score": 0.7076256069009901,
636
- "source": "pentatonic-memory",
637
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
638
- },
639
- {
640
- "text": "billing.new_invoice_pdf",
641
- "score": 0.6999989250830485,
642
- "source": "pentatonic-memory",
643
- "doc_id": "mem_ca8bdac3-12ca-4bae-b5a2-fb9c2d069173"
644
- },
645
- {
646
- "text": "Stripe posts a invoice.payment_succeeded webhook to billing-svc",
647
- "score": 0.6846413593936166,
648
- "source": "pentatonic-memory",
649
- "doc_id": "mem_76c60bf6-9327-4cd1-9f56-234d229e8c8a"
650
- }
651
- ],
652
- "correct": false,
653
- "score": 0.0,
654
- "grading_notes": "missing 1/1: ['launched']",
655
- "search_time_ms": 35.42196500347927,
656
- "generation_time_ms": 0.0,
657
- "tokens_in": 0,
658
- "tokens_out": 0,
659
- "retrieval_tokens": 44,
660
- "query_tokens": 9,
661
- "context_tokens": 35,
662
- "judge_tokens_in": 0,
663
- "judge_tokens_out": 0,
664
- "judge_latency_ms": 0.0
665
- },
666
- {
667
- "task_id": "multihop-atlas",
668
- "query": "Who owns Atlas and what is its current deadline?",
669
- "answer": "The Atlas project is currently in development.\n---\nIf secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nThe user's latest Cirrus deadline is April 8, 2026.",
670
- "hits": [
671
- {
672
- "text": "The Atlas project is currently in development.",
673
- "score": 0.7401179380365475,
674
- "source": "pentatonic-memory",
675
- "doc_id": "chat-deadline2-atlas"
676
- },
677
- {
678
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
679
- "score": 0.7272517964813692,
680
- "source": "pentatonic-memory",
681
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
682
- },
683
- {
684
- "text": "The user's latest Cirrus deadline is April 8, 2026.",
685
- "score": 0.7197845468232443,
686
- "source": "pentatonic-memory",
687
- "doc_id": "chat-deadline2-cirrus"
688
- }
689
- ],
690
- "correct": false,
691
- "score": 0.0,
692
- "grading_notes": "missing 2/2: ['Alice', '2026-03-17']",
693
- "search_time_ms": 35.946914984378964,
694
- "generation_time_ms": 0.0,
695
- "tokens_in": 0,
696
- "tokens_out": 0,
697
- "retrieval_tokens": 52,
698
- "query_tokens": 10,
699
- "context_tokens": 42,
700
- "judge_tokens_in": 0,
701
- "judge_tokens_out": 0,
702
- "judge_latency_ms": 0.0
703
- },
704
- {
705
- "task_id": "multihop-borealis",
706
- "query": "Who owns Borealis and what is its current deadline?",
707
- "answer": "The user's latest Cirrus deadline is April 8, 2026.\n---\nIf secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nhandle_webhook verifies signature, enqueues internal InvoicePaid event on order-svc",
708
- "hits": [
709
- {
710
- "text": "The user's latest Cirrus deadline is April 8, 2026.",
711
- "score": 0.7312607039203609,
712
- "source": "pentatonic-memory",
713
- "doc_id": "chat-deadline2-cirrus"
714
- },
715
- {
716
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
717
- "score": 0.7117345215943829,
718
- "source": "pentatonic-memory",
719
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
720
- },
721
- {
722
- "text": "handle_webhook verifies signature, enqueues internal InvoicePaid event on order-svc",
723
- "score": 0.7061180599052448,
724
- "source": "pentatonic-memory",
725
- "doc_id": "mem_b6596977-9a91-4469-a8c5-3872edbbbf1c"
726
- }
727
- ],
728
- "correct": false,
729
- "score": 0.0,
730
- "grading_notes": "missing 2/2: ['Clara', '2026-02-28']",
731
- "search_time_ms": 36.55060098390095,
732
- "generation_time_ms": 0.0,
733
- "tokens_in": 0,
734
- "tokens_out": 0,
735
- "retrieval_tokens": 62,
736
- "query_tokens": 12,
737
- "context_tokens": 50,
738
- "judge_tokens_in": 0,
739
- "judge_tokens_out": 0,
740
- "judge_latency_ms": 0.0
741
- },
742
- {
743
- "task_id": "multihop-cirrus",
744
- "query": "Who owns Cirrus and what is its current deadline?",
745
- "answer": "The user's latest Cirrus deadline is April 8, 2026.\n---\nIf secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nhandle_webhook verifies signature, enqueues internal InvoicePaid event on order-svc",
746
- "hits": [
747
- {
748
- "text": "The user's latest Cirrus deadline is April 8, 2026.",
749
- "score": 0.7949980719613424,
750
- "source": "pentatonic-memory",
751
- "doc_id": "chat-deadline2-cirrus"
752
- },
753
- {
754
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
755
- "score": 0.7244226449863047,
756
- "source": "pentatonic-memory",
757
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
758
- },
759
- {
760
- "text": "handle_webhook verifies signature, enqueues internal InvoicePaid event on order-svc",
761
- "score": 0.7119401034180405,
762
- "source": "pentatonic-memory",
763
- "doc_id": "mem_b6596977-9a91-4469-a8c5-3872edbbbf1c"
764
- }
765
- ],
766
- "correct": false,
767
- "score": 0.0,
768
- "grading_notes": "missing 2/2: ['Diego', '2026-04-08']",
769
- "search_time_ms": 35.7303679920733,
770
- "generation_time_ms": 0.0,
771
- "tokens_in": 0,
772
- "tokens_out": 0,
773
- "retrieval_tokens": 61,
774
- "query_tokens": 11,
775
- "context_tokens": 50,
776
- "judge_tokens_in": 0,
777
- "judge_tokens_out": 0,
778
- "judge_latency_ms": 0.0
779
- },
780
- {
781
- "task_id": "multihop-dune",
782
- "query": "Who owns Dune and what is its current deadline?",
783
- "answer": "The user's latest Cirrus deadline is April 8, 2026.\n---\nIf secret was rotated recently, `kubectl rollout restart deployment/billing-svc`\n---\nbilling.new_invoice_pdf",
784
- "hits": [
785
- {
786
- "text": "The user's latest Cirrus deadline is April 8, 2026.",
787
- "score": 0.7311771760384164,
788
- "source": "pentatonic-memory",
789
- "doc_id": "chat-deadline2-cirrus"
790
- },
791
- {
792
- "text": "If secret was rotated recently, `kubectl rollout restart deployment/billing-svc`",
793
- "score": 0.7118296546980413,
794
- "source": "pentatonic-memory",
795
- "doc_id": "mem_bb06d70a-c2f5-4168-b0df-a03b1b585b29"
796
- },
797
- {
798
- "text": "billing.new_invoice_pdf",
799
- "score": 0.6992845290215867,
800
- "source": "pentatonic-memory",
801
- "doc_id": "mem_ca8bdac3-12ca-4bae-b5a2-fb9c2d069173"
802
- }
803
- ],
804
- "correct": false,
805
- "score": 0.0,
806
- "grading_notes": "missing 2/2: ['Farid', '2026-05-20']",
807
- "search_time_ms": 35.289291001390666,
808
- "generation_time_ms": 0.0,
809
- "tokens_in": 0,
810
- "tokens_out": 0,
811
- "retrieval_tokens": 49,
812
- "query_tokens": 11,
813
- "context_tokens": 38,
814
- "judge_tokens_in": 0,
815
- "judge_tokens_out": 0,
816
- "judge_latency_ms": 0.0
817
- }
818
- ]
819
- }