open-research-protocol 0.4.6 → 0.4.8

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 (46) hide show
  1. package/README.md +10 -0
  2. package/cli/orp.py +668 -43
  3. package/docs/ORP_REASONING_KERNEL_AGENT_PILOT.md +125 -0
  4. package/docs/ORP_REASONING_KERNEL_AGENT_REPLICATION.md +97 -0
  5. package/docs/ORP_REASONING_KERNEL_CANONICAL_CONTINUATION_PILOT.md +100 -0
  6. package/docs/ORP_REASONING_KERNEL_COMPARISON_PILOT.md +116 -0
  7. package/docs/ORP_REASONING_KERNEL_CONTINUATION_PILOT.md +86 -0
  8. package/docs/ORP_REASONING_KERNEL_EVALUATION_PLAN.md +261 -0
  9. package/docs/ORP_REASONING_KERNEL_EVIDENCE_MATRIX.md +131 -0
  10. package/docs/ORP_REASONING_KERNEL_EVOLUTION.md +123 -0
  11. package/docs/ORP_REASONING_KERNEL_PICKUP_PILOT.md +107 -0
  12. package/docs/ORP_REASONING_KERNEL_TECHNICAL_VALIDATION.md +471 -0
  13. package/docs/ORP_REASONING_KERNEL_V0_1.md +15 -0
  14. package/docs/benchmarks/orp_reasoning_kernel_agent_pilot_v0_1.json +796 -0
  15. package/docs/benchmarks/orp_reasoning_kernel_agent_replication_task_smoke.json +487 -0
  16. package/docs/benchmarks/orp_reasoning_kernel_agent_replication_v0_1.json +1927 -0
  17. package/docs/benchmarks/orp_reasoning_kernel_agent_replication_v0_2.json +10217 -0
  18. package/docs/benchmarks/orp_reasoning_kernel_canonical_continuation_task_smoke.json +174 -0
  19. package/docs/benchmarks/orp_reasoning_kernel_canonical_continuation_v0_1.json +598 -0
  20. package/docs/benchmarks/orp_reasoning_kernel_comparison_v0_1.json +688 -0
  21. package/docs/benchmarks/orp_reasoning_kernel_continuation_task_smoke.json +150 -0
  22. package/docs/benchmarks/orp_reasoning_kernel_continuation_v0_1.json +448 -0
  23. package/docs/benchmarks/orp_reasoning_kernel_pickup_v0_1.json +594 -0
  24. package/docs/benchmarks/orp_reasoning_kernel_v0_1_validation.json +925 -0
  25. package/examples/README.md +2 -0
  26. package/examples/kernel/comparison/comparison-corpus.json +337 -0
  27. package/examples/kernel/comparison/next-task-continuation.json +55 -0
  28. package/examples/kernel/corpus/operations/habanero-routing.checkpoint.kernel.yml +12 -0
  29. package/examples/kernel/corpus/operations/runner-routing.policy.kernel.yml +9 -0
  30. package/examples/kernel/corpus/product/project-home.decision.kernel.yml +11 -0
  31. package/examples/kernel/corpus/research/kernel-handoff.experiment.kernel.yml +16 -0
  32. package/examples/kernel/corpus/research/lane-drift.hypothesis.kernel.yml +11 -0
  33. package/examples/kernel/corpus/software/trace-widget.task.kernel.yml +13 -0
  34. package/examples/kernel/corpus/writing/kernel-launch.result.kernel.yml +12 -0
  35. package/package.json +4 -1
  36. package/scripts/orp-kernel-agent-pilot.py +673 -0
  37. package/scripts/orp-kernel-agent-replication.py +307 -0
  38. package/scripts/orp-kernel-benchmark.py +921 -0
  39. package/scripts/orp-kernel-canonical-continuation.py +381 -0
  40. package/scripts/orp-kernel-ci-check.py +138 -0
  41. package/scripts/orp-kernel-comparison.py +592 -0
  42. package/scripts/orp-kernel-continuation-pilot.py +384 -0
  43. package/scripts/orp-kernel-pickup.py +401 -0
  44. package/spec/v1/kernel-extension.schema.json +96 -0
  45. package/spec/v1/kernel-proposal.schema.json +115 -0
  46. package/spec/v1/kernel.schema.json +2 -1
package/docs/benchmarks/orp_reasoning_kernel_agent_replication_task_smoke.json ADDED
@@ -0,0 +1,487 @@
1
+ {
2
+ "schema_version": "1.0.0",
3
+ "kind": "orp_reasoning_kernel_agent_replication_report",
4
+ "metadata": {
5
+ "generated_at_utc": "2026-03-23T07:53:55Z",
6
+ "repo_commit": "c2f7f2a52744a00fb719d37de583da1f4ae615bd",
7
+ "repo_branch": "main",
8
+ "package_version": "0.4.7",
9
+ "python_version": "3.9.6",
10
+ "codex_version": "codex-cli 0.116.0",
11
+ "platform": "macOS-26.3-arm64-arm-64bit",
12
+ "model": "default",
13
+ "repeats": 2
14
+ },
15
+ "runs": [
16
+ {
17
+ "run_index": 1,
18
+ "summary": {
19
+ "all_claims_pass": true,
20
+ "kernel_mean_pickup_score": 1.0,
21
+ "generic_checklist_mean_pickup_score": 0.8,
22
+ "freeform_mean_pickup_score": 0.6,
23
+ "kernel_mean_invention_rate": 0.0,
24
+ "generic_checklist_mean_invention_rate": 0.0,
25
+ "freeform_mean_invention_rate": 0.0
26
+ },
27
+ "conditions": {
28
+ "freeform": {
29
+ "condition": "freeform",
30
+ "cases_total": 1,
31
+ "rows": [
32
+ {
33
+ "id": "software_trace_widget",
34
+ "domain": "software",
35
+ "artifact_class": "task",
36
+ "pickup_score": 0.6,
37
+ "ambiguity_remaining": 0.4,
38
+ "answered_targets": 3,
39
+ "pickup_targets_total": 5,
40
+ "expected_present_fields": [
41
+ "constraints",
42
+ "goal",
43
+ "object"
44
+ ],
45
+ "answers": {
46
+ "object": "terminal trace widget for lane monitoring",
47
+ "goal": "let operators tell quickly when a lane is drifting",
48
+ "boundary": null,
49
+ "constraints": "stay terminal-first and low friction",
50
+ "success_criteria": null
51
+ },
52
+ "invented_fields": [],
53
+ "invention_rate": 0.0,
54
+ "artifact_type_guess": "lightweight design brief",
55
+ "confidence": 0.97,
56
+ "ambiguities_count": 2,
57
+ "elapsed_ms": 15054.326,
58
+ "tokens_used": null,
59
+ "session_id": ""
60
+ }
61
+ ],
62
+ "mean_pickup_score": 0.6,
63
+ "mean_ambiguity_remaining": 0.4,
64
+ "mean_answered_target_rate": 0.6,
65
+ "mean_confidence": 0.97,
66
+ "mean_ambiguities_count": 2.0,
67
+ "mean_invention_rate": 0.0,
68
+ "mean_elapsed_ms": 15054.326,
69
+ "mean_tokens_used": null
70
+ },
71
+ "generic_checklist": {
72
+ "condition": "generic_checklist",
73
+ "cases_total": 1,
74
+ "rows": [
75
+ {
76
+ "id": "software_trace_widget",
77
+ "domain": "software",
78
+ "artifact_class": "task",
79
+ "pickup_score": 0.8,
80
+ "ambiguity_remaining": 0.2,
81
+ "answered_targets": 4,
82
+ "pickup_targets_total": 5,
83
+ "expected_present_fields": [
84
+ "boundary",
85
+ "constraints",
86
+ "goal",
87
+ "object",
88
+ "success_criteria"
89
+ ],
90
+ "answers": {
91
+ "object": null,
92
+ "goal": "Build the terminal trace widget for lane monitoring.",
93
+ "boundary": "Terminal-first lane visibility in active ORP sessions.",
94
+ "constraints": "[\"low friction\", \"no GUI dependency\"]",
95
+ "success_criteria": "[\"An operator can identify a drifting lane quickly.\"]"
96
+ },
97
+ "invented_fields": [],
98
+ "invention_rate": 0.0,
99
+ "artifact_type_guess": "task",
100
+ "confidence": 0.84,
101
+ "ambiguities_count": 2,
102
+ "elapsed_ms": 62306.677,
103
+ "tokens_used": null,
104
+ "session_id": ""
105
+ }
106
+ ],
107
+ "mean_pickup_score": 0.8,
108
+ "mean_ambiguity_remaining": 0.2,
109
+ "mean_answered_target_rate": 0.8,
110
+ "mean_confidence": 0.84,
111
+ "mean_ambiguities_count": 2.0,
112
+ "mean_invention_rate": 0.0,
113
+ "mean_elapsed_ms": 62306.677,
114
+ "mean_tokens_used": null
115
+ },
116
+ "kernel": {
117
+ "condition": "kernel",
118
+ "cases_total": 1,
119
+ "rows": [
120
+ {
121
+ "id": "software_trace_widget",
122
+ "domain": "software",
123
+ "artifact_class": "task",
124
+ "pickup_score": 1.0,
125
+ "ambiguity_remaining": 0.0,
126
+ "answered_targets": 5,
127
+ "pickup_targets_total": 5,
128
+ "expected_present_fields": [
129
+ "boundary",
130
+ "constraints",
131
+ "goal",
132
+ "object",
133
+ "success_criteria"
134
+ ],
135
+ "answers": {
136
+ "object": "terminal trace widget",
137
+ "goal": "surface lane drift and state clearly for operators",
138
+ "boundary": "[\"terminal-first lane visibility\",\"active ORP sessions only\"]",
139
+ "constraints": "[\"low friction\",\"no GUI dependency\"]",
140
+ "success_criteria": "[\"an operator can identify a drifting lane within 10 seconds\",\"the widget does not overload the terminal surface\"]"
141
+ },
142
+ "invented_fields": [],
143
+ "invention_rate": 0.0,
144
+ "artifact_type_guess": "task",
145
+ "confidence": 0.99,
146
+ "ambiguities_count": 0,
147
+ "elapsed_ms": 26515.229,
148
+ "tokens_used": null,
149
+ "session_id": ""
150
+ }
151
+ ],
152
+ "mean_pickup_score": 1.0,
153
+ "mean_ambiguity_remaining": 0.0,
154
+ "mean_answered_target_rate": 1.0,
155
+ "mean_confidence": 0.99,
156
+ "mean_ambiguities_count": 0.0,
157
+ "mean_invention_rate": 0.0,
158
+ "mean_elapsed_ms": 26515.229,
159
+ "mean_tokens_used": null
160
+ }
161
+ },
162
+ "pairwise": {
163
+ "kernel_vs_generic_checklist": {
164
+ "left": "kernel",
165
+ "right": "generic_checklist",
166
+ "wins": 1,
167
+ "ties": 0,
168
+ "losses": 0,
169
+ "mean_pickup_score_delta": 0.2,
170
+ "by_case": [
171
+ {
172
+ "id": "software_trace_widget",
173
+ "domain": "software",
174
+ "artifact_class": "task",
175
+ "left_score": 1.0,
176
+ "right_score": 0.8,
177
+ "delta": 0.2,
178
+ "outcome": "win"
179
+ }
180
+ ]
181
+ },
182
+ "kernel_vs_freeform": {
183
+ "left": "kernel",
184
+ "right": "freeform",
185
+ "wins": 1,
186
+ "ties": 0,
187
+ "losses": 0,
188
+ "mean_pickup_score_delta": 0.4,
189
+ "by_case": [
190
+ {
191
+ "id": "software_trace_widget",
192
+ "domain": "software",
193
+ "artifact_class": "task",
194
+ "left_score": 1.0,
195
+ "right_score": 0.6,
196
+ "delta": 0.4,
197
+ "outcome": "win"
198
+ }
199
+ ]
200
+ },
201
+ "generic_checklist_vs_freeform": {
202
+ "left": "generic_checklist",
203
+ "right": "freeform",
204
+ "wins": 1,
205
+ "ties": 0,
206
+ "losses": 0,
207
+ "mean_pickup_score_delta": 0.2,
208
+ "by_case": [
209
+ {
210
+ "id": "software_trace_widget",
211
+ "domain": "software",
212
+ "artifact_class": "task",
213
+ "left_score": 0.8,
214
+ "right_score": 0.6,
215
+ "delta": 0.2,
216
+ "outcome": "win"
217
+ }
218
+ ]
219
+ }
220
+ }
221
+ },
222
+ {
223
+ "run_index": 2,
224
+ "summary": {
225
+ "all_claims_pass": false,
226
+ "kernel_mean_pickup_score": 1.0,
227
+ "generic_checklist_mean_pickup_score": 1.0,
228
+ "freeform_mean_pickup_score": 0.6,
229
+ "kernel_mean_invention_rate": 0.0,
230
+ "generic_checklist_mean_invention_rate": 0.0,
231
+ "freeform_mean_invention_rate": 0.0
232
+ },
233
+ "conditions": {
234
+ "freeform": {
235
+ "condition": "freeform",
236
+ "cases_total": 1,
237
+ "rows": [
238
+ {
239
+ "id": "software_trace_widget",
240
+ "domain": "software",
241
+ "artifact_class": "task",
242
+ "pickup_score": 0.6,
243
+ "ambiguity_remaining": 0.4,
244
+ "answered_targets": 3,
245
+ "pickup_targets_total": 5,
246
+ "expected_present_fields": [
247
+ "constraints",
248
+ "goal",
249
+ "object"
250
+ ],
251
+ "answers": {
252
+ "object": "terminal trace widget for lane monitoring.",
253
+ "goal": "let operators tell quickly when a lane is drifting.",
254
+ "boundary": null,
255
+ "constraints": "stay terminal-first and low friction.",
256
+ "success_criteria": null
257
+ },
258
+ "invented_fields": [],
259
+ "invention_rate": 0.0,
260
+ "artifact_type_guess": "design brief",
261
+ "confidence": 0.94,
262
+ "ambiguities_count": 2,
263
+ "elapsed_ms": 15578.968,
264
+ "tokens_used": null,
265
+ "session_id": ""
266
+ }
267
+ ],
268
+ "mean_pickup_score": 0.6,
269
+ "mean_ambiguity_remaining": 0.4,
270
+ "mean_answered_target_rate": 0.6,
271
+ "mean_confidence": 0.94,
272
+ "mean_ambiguities_count": 2.0,
273
+ "mean_invention_rate": 0.0,
274
+ "mean_elapsed_ms": 15578.968,
275
+ "mean_tokens_used": null
276
+ },
277
+ "generic_checklist": {
278
+ "condition": "generic_checklist",
279
+ "cases_total": 1,
280
+ "rows": [
281
+ {
282
+ "id": "software_trace_widget",
283
+ "domain": "software",
284
+ "artifact_class": "task",
285
+ "pickup_score": 1.0,
286
+ "ambiguity_remaining": 0.0,
287
+ "answered_targets": 5,
288
+ "pickup_targets_total": 5,
289
+ "expected_present_fields": [
290
+ "boundary",
291
+ "constraints",
292
+ "goal",
293
+ "object",
294
+ "success_criteria"
295
+ ],
296
+ "answers": {
297
+ "object": "terminal trace widget for lane monitoring",
298
+ "goal": "Build the terminal trace widget for lane monitoring.",
299
+ "boundary": "Terminal-first lane visibility in active ORP sessions.",
300
+ "constraints": "low friction; no GUI dependency",
301
+ "success_criteria": "An operator can identify a drifting lane quickly."
302
+ },
303
+ "invented_fields": [],
304
+ "invention_rate": 0.0,
305
+ "artifact_type_guess": "task",
306
+ "confidence": 0.86,
307
+ "ambiguities_count": 2,
308
+ "elapsed_ms": 39661.065,
309
+ "tokens_used": null,
310
+ "session_id": ""
311
+ }
312
+ ],
313
+ "mean_pickup_score": 1.0,
314
+ "mean_ambiguity_remaining": 0.0,
315
+ "mean_answered_target_rate": 1.0,
316
+ "mean_confidence": 0.86,
317
+ "mean_ambiguities_count": 2.0,
318
+ "mean_invention_rate": 0.0,
319
+ "mean_elapsed_ms": 39661.065,
320
+ "mean_tokens_used": null
321
+ },
322
+ "kernel": {
323
+ "condition": "kernel",
324
+ "cases_total": 1,
325
+ "rows": [
326
+ {
327
+ "id": "software_trace_widget",
328
+ "domain": "software",
329
+ "artifact_class": "task",
330
+ "pickup_score": 1.0,
331
+ "ambiguity_remaining": 0.0,
332
+ "answered_targets": 5,
333
+ "pickup_targets_total": 5,
334
+ "expected_present_fields": [
335
+ "boundary",
336
+ "constraints",
337
+ "goal",
338
+ "object",
339
+ "success_criteria"
340
+ ],
341
+ "answers": {
342
+ "object": "terminal trace widget",
343
+ "goal": "surface lane drift and state clearly for operators",
344
+ "boundary": "[\"terminal-first lane visibility\",\"active ORP sessions only\"]",
345
+ "constraints": "[\"low friction\",\"no GUI dependency\"]",
346
+ "success_criteria": "[\"an operator can identify a drifting lane within 10 seconds\",\"the widget does not overload the terminal surface\"]"
347
+ },
348
+ "invented_fields": [],
349
+ "invention_rate": 0.0,
350
+ "artifact_type_guess": "task",
351
+ "confidence": 0.99,
352
+ "ambiguities_count": 0,
353
+ "elapsed_ms": 26937.578,
354
+ "tokens_used": null,
355
+ "session_id": ""
356
+ }
357
+ ],
358
+ "mean_pickup_score": 1.0,
359
+ "mean_ambiguity_remaining": 0.0,
360
+ "mean_answered_target_rate": 1.0,
361
+ "mean_confidence": 0.99,
362
+ "mean_ambiguities_count": 0.0,
363
+ "mean_invention_rate": 0.0,
364
+ "mean_elapsed_ms": 26937.578,
365
+ "mean_tokens_used": null
366
+ }
367
+ },
368
+ "pairwise": {
369
+ "kernel_vs_generic_checklist": {
370
+ "left": "kernel",
371
+ "right": "generic_checklist",
372
+ "wins": 0,
373
+ "ties": 1,
374
+ "losses": 0,
375
+ "mean_pickup_score_delta": 0.0,
376
+ "by_case": [
377
+ {
378
+ "id": "software_trace_widget",
379
+ "domain": "software",
380
+ "artifact_class": "task",
381
+ "left_score": 1.0,
382
+ "right_score": 1.0,
383
+ "delta": 0.0,
384
+ "outcome": "tie"
385
+ }
386
+ ]
387
+ },
388
+ "kernel_vs_freeform": {
389
+ "left": "kernel",
390
+ "right": "freeform",
391
+ "wins": 1,
392
+ "ties": 0,
393
+ "losses": 0,
394
+ "mean_pickup_score_delta": 0.4,
395
+ "by_case": [
396
+ {
397
+ "id": "software_trace_widget",
398
+ "domain": "software",
399
+ "artifact_class": "task",
400
+ "left_score": 1.0,
401
+ "right_score": 0.6,
402
+ "delta": 0.4,
403
+ "outcome": "win"
404
+ }
405
+ ]
406
+ },
407
+ "generic_checklist_vs_freeform": {
408
+ "left": "generic_checklist",
409
+ "right": "freeform",
410
+ "wins": 1,
411
+ "ties": 0,
412
+ "losses": 0,
413
+ "mean_pickup_score_delta": 0.4,
414
+ "by_case": [
415
+ {
416
+ "id": "software_trace_widget",
417
+ "domain": "software",
418
+ "artifact_class": "task",
419
+ "left_score": 1.0,
420
+ "right_score": 0.6,
421
+ "delta": 0.4,
422
+ "outcome": "win"
423
+ }
424
+ ]
425
+ }
426
+ }
427
+ }
428
+ ],
429
+ "conditions": {
430
+ "freeform": {
431
+ "mean_pickup_score": 0.6,
432
+ "pickup_score_stdev": 0.0,
433
+ "mean_invention_rate": 0.0,
434
+ "invention_rate_stdev": 0.0,
435
+ "mean_confidence": 0.955,
436
+ "confidence_stdev": 0.015,
437
+ "mean_elapsed_ms": 15316.647,
438
+ "elapsed_ms_stdev": 262.321
439
+ },
440
+ "generic_checklist": {
441
+ "mean_pickup_score": 0.9,
442
+ "pickup_score_stdev": 0.1,
443
+ "mean_invention_rate": 0.0,
444
+ "invention_rate_stdev": 0.0,
445
+ "mean_confidence": 0.85,
446
+ "confidence_stdev": 0.01,
447
+ "mean_elapsed_ms": 50983.871,
448
+ "elapsed_ms_stdev": 11322.806
449
+ },
450
+ "kernel": {
451
+ "mean_pickup_score": 1.0,
452
+ "pickup_score_stdev": 0.0,
453
+ "mean_invention_rate": 0.0,
454
+ "invention_rate_stdev": 0.0,
455
+ "mean_confidence": 0.99,
456
+ "confidence_stdev": 0.0,
457
+ "mean_elapsed_ms": 26726.404,
458
+ "elapsed_ms_stdev": 211.175
459
+ }
460
+ },
461
+ "claims": [
462
+ {
463
+ "id": "kernel_stays_above_generic_checklist_across_replication",
464
+ "claim": "Across repeated live Codex runs, kernel mean pickup stays at or above generic checklist mean pickup, and above it on the aggregated sample.",
465
+ "status": "pass"
466
+ },
467
+ {
468
+ "id": "kernel_stays_above_freeform_across_replication",
469
+ "claim": "Across repeated live Codex runs, kernel mean pickup stays above free-form mean pickup.",
470
+ "status": "pass"
471
+ },
472
+ {
473
+ "id": "kernel_keeps_lowest_or_equal_invention_rate_across_replication",
474
+ "claim": "Across repeated live Codex runs, kernel mean invention rate stays at or below the other conditions.",
475
+ "status": "pass"
476
+ }
477
+ ],
478
+ "summary": {
479
+ "all_claims_pass": true,
480
+ "kernel_mean_pickup_score": 1.0,
481
+ "generic_checklist_mean_pickup_score": 0.9,
482
+ "freeform_mean_pickup_score": 0.6,
483
+ "kernel_mean_invention_rate": 0.0,
484
+ "generic_checklist_mean_invention_rate": 0.0,
485
+ "freeform_mean_invention_rate": 0.0
486
+ }
487
+ }