@riddledc/riddle-proof 0.5.31 → 0.5.33
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.
- package/dist/{chunk-R4BPJNAM.js → chunk-A5AWVY5A.js} +81 -6
- package/dist/{chunk-A3H7GU7H.js → chunk-CHKYSZLU.js} +39 -1
- package/dist/engine-harness.cjs +120 -9
- package/dist/engine-harness.js +2 -1
- package/dist/index.cjs +120 -9
- package/dist/index.js +2 -1
- package/dist/proof-run-core.cjs +84 -6
- package/dist/proof-run-core.d.cts +8 -1
- package/dist/proof-run-core.d.ts +8 -1
- package/dist/proof-run-core.js +7 -1
- package/dist/proof-run-engine.cjs +82 -7
- package/dist/proof-run-engine.js +5 -2
- package/package.json +1 -1
- package/runtime/lib/ship.py +55 -1
- package/runtime/lib/verify.py +38 -4
- package/runtime/tests/recon_verify_smoke.py +61 -2
package/dist/proof-run-engine.js
CHANGED
|
@@ -14,7 +14,7 @@ import {
|
|
|
14
14
|
validateShipGate,
|
|
15
15
|
workflowFile,
|
|
16
16
|
writeState
|
|
17
|
-
} from "./chunk-
|
|
17
|
+
} from "./chunk-A5AWVY5A.js";
|
|
18
18
|
|
|
19
19
|
// src/proof-run-engine.ts
|
|
20
20
|
import { execFileSync } from "child_process";
|
|
@@ -770,6 +770,9 @@ async function executeWorkflow(params, pluginConfig, resolvedConfig) {
|
|
|
770
770
|
if (reasons.some((reason) => reason.includes("proof_assessment"))) {
|
|
771
771
|
return "resume with riddle_proof_review using decision=ready_to_ship only after the screenshots and semantic evidence visibly prove the request; otherwise choose needs_implementation or needs_richer_proof";
|
|
772
772
|
}
|
|
773
|
+
if (reasons.some((reason) => reason.includes("visual_delta"))) {
|
|
774
|
+
return "rerun verify with a measured before/after visual delta, or choose needs_richer_proof instead of ready_to_ship for this visual proof";
|
|
775
|
+
}
|
|
773
776
|
if (reasons.some((reason) => reason.includes("after_cdn") || reason.includes("verify_status"))) {
|
|
774
777
|
return "rerun verify with stronger proof framing so after evidence is captured before shipping";
|
|
775
778
|
}
|
|
@@ -862,7 +865,7 @@ async function executeWorkflow(params, pluginConfig, resolvedConfig) {
|
|
|
862
865
|
}
|
|
863
866
|
state = readState(config.statePath);
|
|
864
867
|
const continuedStage = params.continue_from_checkpoint ? checkpointContinueStage(state) : null;
|
|
865
|
-
if (params.continue_from_checkpoint && !continuedStage) {
|
|
868
|
+
if (params.continue_from_checkpoint && !params.advance_stage && !continuedStage) {
|
|
866
869
|
const recommended = recommendedAdvanceStage(state);
|
|
867
870
|
return checkpoint(
|
|
868
871
|
state?.active_checkpoint_stage || recommended || "recon",
|
package/package.json
CHANGED
package/runtime/lib/ship.py
CHANGED
|
@@ -9,6 +9,10 @@ from util import load_state, save_state, invoke, git
|
|
|
9
9
|
|
|
10
10
|
DISCORD_API = 'https://discord.com/api/v10'
|
|
11
11
|
SHIP_NOISE_PATHS = ('.codex', '.oc-smoke')
|
|
12
|
+
VISUAL_FIRST_MODES = {
|
|
13
|
+
'visual', 'render', 'interaction', 'ui', 'layout', 'screenshot',
|
|
14
|
+
'canvas', 'animation',
|
|
15
|
+
}
|
|
12
16
|
|
|
13
17
|
|
|
14
18
|
def read_json_file(path):
|
|
@@ -293,7 +297,11 @@ def record_ship_report(state, marked_ready=None):
|
|
|
293
297
|
def proof_assessment_is_ready(state):
|
|
294
298
|
assessment = state.get('proof_assessment') or {}
|
|
295
299
|
source = str(assessment.get('source') or state.get('proof_assessment_source') or '').strip().lower()
|
|
296
|
-
return
|
|
300
|
+
return (
|
|
301
|
+
source in ('supervising_agent', 'supervisor')
|
|
302
|
+
and assessment.get('decision') == 'ready_to_ship'
|
|
303
|
+
and not visual_delta_ship_blocker(state)
|
|
304
|
+
)
|
|
297
305
|
|
|
298
306
|
|
|
299
307
|
def effective_merge_recommendation(state):
|
|
@@ -310,6 +318,49 @@ def after_evidence_bundle(state):
|
|
|
310
318
|
return after if isinstance(after, dict) else {}
|
|
311
319
|
|
|
312
320
|
|
|
321
|
+
def normalized_verification_mode(state):
|
|
322
|
+
bundle = state.get('evidence_bundle') or {}
|
|
323
|
+
if isinstance(bundle, dict) and str(bundle.get('verification_mode') or '').strip():
|
|
324
|
+
return str(bundle.get('verification_mode')).strip().lower()
|
|
325
|
+
return str(state.get('verification_mode') or 'proof').strip().lower() or 'proof'
|
|
326
|
+
|
|
327
|
+
|
|
328
|
+
def visual_delta_required_for_ship(state):
|
|
329
|
+
bundle = state.get('evidence_bundle') or {}
|
|
330
|
+
contract = bundle.get('artifact_contract') if isinstance(bundle, dict) else {}
|
|
331
|
+
required = contract.get('required') if isinstance(contract, dict) else {}
|
|
332
|
+
if isinstance(required, dict) and required.get('visual_delta') is True:
|
|
333
|
+
return True
|
|
334
|
+
return normalized_verification_mode(state) in VISUAL_FIRST_MODES
|
|
335
|
+
|
|
336
|
+
|
|
337
|
+
def visual_delta_for_state(state):
|
|
338
|
+
after = after_evidence_bundle(state)
|
|
339
|
+
visual_delta = after.get('visual_delta') if isinstance(after, dict) else None
|
|
340
|
+
if isinstance(visual_delta, dict):
|
|
341
|
+
return visual_delta
|
|
342
|
+
request = state.get('proof_assessment_request') or {}
|
|
343
|
+
visual_delta = request.get('visual_delta') if isinstance(request, dict) else None
|
|
344
|
+
return visual_delta if isinstance(visual_delta, dict) else {}
|
|
345
|
+
|
|
346
|
+
|
|
347
|
+
def visual_delta_ship_blocker(state):
|
|
348
|
+
if not visual_delta_required_for_ship(state):
|
|
349
|
+
return ''
|
|
350
|
+
visual_delta = visual_delta_for_state(state)
|
|
351
|
+
if visual_delta.get('status') == 'measured' and visual_delta.get('passed') is True:
|
|
352
|
+
return ''
|
|
353
|
+
status = str(visual_delta.get('status') or 'missing')
|
|
354
|
+
if status == 'unmeasured':
|
|
355
|
+
return 'visual_delta.status=unmeasured blocks ready_to_ship for visual/UI proof'
|
|
356
|
+
if status == 'measured' and visual_delta.get('passed') is False:
|
|
357
|
+
return 'visual_delta.status=measured but visual_delta.passed=false blocks ready_to_ship for visual/UI proof'
|
|
358
|
+
reason = str(visual_delta.get('reason') or '').strip()
|
|
359
|
+
if reason:
|
|
360
|
+
return f'visual_delta.status={status} blocks ready_to_ship for visual/UI proof: {reason}'
|
|
361
|
+
return f'visual_delta.status={status} blocks ready_to_ship for visual/UI proof'
|
|
362
|
+
|
|
363
|
+
|
|
313
364
|
def state_has_after_evidence(state):
|
|
314
365
|
if (state.get('after_cdn') or '').strip():
|
|
315
366
|
return True
|
|
@@ -581,6 +632,9 @@ if reference in ('prod', 'both'):
|
|
|
581
632
|
raise SystemExit('prod_url is required when reference=' + reference + ' before ship.')
|
|
582
633
|
if not prod_cdn:
|
|
583
634
|
raise SystemExit('prod_cdn is required before ship. Run recon/verify again and preserve the approved prod baseline.')
|
|
635
|
+
visual_delta_blocker = visual_delta_ship_blocker(s)
|
|
636
|
+
if visual_delta_blocker:
|
|
637
|
+
raise SystemExit(visual_delta_blocker + '. Rerun verify with measured before/after visual delta or return a non-shipping proof assessment.')
|
|
584
638
|
if proof_source not in ('supervising_agent', 'supervisor') or proof_assessment.get('decision') != 'ready_to_ship':
|
|
585
639
|
raise SystemExit('Supervising-agent proof_assessment.decision=ready_to_ship is required before ship.')
|
|
586
640
|
|
package/runtime/lib/verify.py
CHANGED
|
@@ -559,6 +559,34 @@ def visual_delta_applies(verification_mode):
|
|
|
559
559
|
return screenshot_required_for_mode(verification_mode)
|
|
560
560
|
|
|
561
561
|
|
|
562
|
+
def visual_delta_passes_ship_gate(visual_delta):
|
|
563
|
+
return (
|
|
564
|
+
isinstance(visual_delta, dict)
|
|
565
|
+
and visual_delta.get('status') == 'measured'
|
|
566
|
+
and visual_delta.get('passed') is True
|
|
567
|
+
)
|
|
568
|
+
|
|
569
|
+
|
|
570
|
+
def visual_delta_blocker_for_mode(verification_mode, visual_delta):
|
|
571
|
+
if not visual_delta_applies(verification_mode):
|
|
572
|
+
return ''
|
|
573
|
+
if visual_delta_passes_ship_gate(visual_delta):
|
|
574
|
+
return ''
|
|
575
|
+
if not isinstance(visual_delta, dict):
|
|
576
|
+
status = 'missing'
|
|
577
|
+
reason = 'No visual_delta object was found in proof evidence.'
|
|
578
|
+
else:
|
|
579
|
+
status = str(visual_delta.get('status') or 'missing')
|
|
580
|
+
reason = str(visual_delta.get('reason') or '').strip()
|
|
581
|
+
if status == 'unmeasured':
|
|
582
|
+
return 'visual_delta.status=unmeasured blocks ready_to_ship for visual/UI proof; capture a measured before/after visual delta or choose needs_richer_proof.'
|
|
583
|
+
if status == 'measured' and isinstance(visual_delta, dict) and visual_delta.get('passed') is False:
|
|
584
|
+
return 'visual_delta.status=measured but visual_delta.passed=false blocks ready_to_ship for visual/UI proof; the measured change did not clear the threshold.'
|
|
585
|
+
if reason:
|
|
586
|
+
return f'visual_delta.status={status} blocks ready_to_ship for visual/UI proof: {reason}'
|
|
587
|
+
return f'visual_delta.status={status} blocks ready_to_ship for visual/UI proof.'
|
|
588
|
+
|
|
589
|
+
|
|
562
590
|
def list_value(value):
|
|
563
591
|
return value if isinstance(value, list) else []
|
|
564
592
|
|
|
@@ -684,11 +712,11 @@ def artifact_contract_for_mode(verification_mode):
|
|
|
684
712
|
'route_semantics': True,
|
|
685
713
|
'screenshot': screenshot_required_for_mode(mode),
|
|
686
714
|
'proof_evidence': proof_evidence_required_for_mode(mode),
|
|
715
|
+
'visual_delta': visual_delta_applies(mode),
|
|
687
716
|
},
|
|
688
717
|
'preferred': {
|
|
689
718
|
'page_state': True,
|
|
690
719
|
'structured_payload': mode in STRUCTURED_FIRST_MODES or proof_evidence_required_for_mode(mode),
|
|
691
|
-
'visual_delta': visual_delta_applies(mode),
|
|
692
720
|
},
|
|
693
721
|
'optional': {
|
|
694
722
|
'console_summary': True,
|
|
@@ -731,7 +759,7 @@ def artifact_signal_availability(state, after_observation, supporting, visual_de
|
|
|
731
759
|
),
|
|
732
760
|
'structured_payload': bool(supporting.get('has_structured_payload')),
|
|
733
761
|
'proof_evidence': bool(supporting.get('proof_evidence_present')),
|
|
734
|
-
'visual_delta':
|
|
762
|
+
'visual_delta': visual_delta_passes_ship_gate(visual_delta),
|
|
735
763
|
'console_summary': bool(supporting.get('console_entries')),
|
|
736
764
|
'json_artifacts': bool(supporting.get('data_outputs')),
|
|
737
765
|
'image_outputs': bool(supporting.get('image_outputs')),
|
|
@@ -1097,6 +1125,8 @@ def build_supervisor_assessment_request(state, payload, after_observation, requi
|
|
|
1097
1125
|
evidence_bundle['artifact_contract'] = artifact_contract
|
|
1098
1126
|
evidence_bundle['artifact_production'] = artifact_production
|
|
1099
1127
|
evidence_bundle['artifact_usage'] = artifact_usage
|
|
1128
|
+
visual_delta_blocker = visual_delta_blocker_for_mode(verification_mode, visual_delta)
|
|
1129
|
+
hard_blockers = [visual_delta_blocker] if visual_delta_blocker else []
|
|
1100
1130
|
|
|
1101
1131
|
return {
|
|
1102
1132
|
'status': 'needs_supervising_agent_assessment',
|
|
@@ -1112,14 +1142,15 @@ def build_supervisor_assessment_request(state, payload, after_observation, requi
|
|
|
1112
1142
|
'artifact_contract': artifact_contract,
|
|
1113
1143
|
'artifact_production': artifact_production,
|
|
1114
1144
|
'artifact_usage': artifact_usage,
|
|
1145
|
+
'hard_blockers': hard_blockers,
|
|
1115
1146
|
'instructions': [
|
|
1116
1147
|
'The supervising agent owns proof assessment. Inspect the recon baseline(s), after evidence, and any structured artifacts together.',
|
|
1117
1148
|
'Decide whether the evidence is ready_to_ship or should continue internally through author, implement, or recon.',
|
|
1149
|
+
'Hard blockers cannot be overridden by supervisor judgment; if hard_blockers is non-empty, do not choose ready_to_ship.',
|
|
1118
1150
|
'Do not mark ready_to_ship if the before/prod baseline is blank, shell-only, generic, or not visibly tied to the requested feature.',
|
|
1119
1151
|
'Use semantic_context.route plus headings/buttons/text anchors to ground route and content judgment before treating a screenshot as wrong-route.',
|
|
1120
1152
|
'For visual/UI modes, use screenshots plus after_observation.details.visible_text_sample, headings, buttons, links, canvas_count, and large_visible_elements to explain what the proof actually shows.',
|
|
1121
|
-
'For visual/UI polish, capture success is not proof. If visual_delta.status
|
|
1122
|
-
'If visual_delta.status=unmeasured for visual/UI proof, only choose ready_to_ship when the screenshots and page-state details let you name a clearly legible before/after change; otherwise request richer proof or another implementation pass.',
|
|
1153
|
+
'For visual/UI polish, capture success is not proof. If visual_delta.status is unmeasured, missing, not_applicable, or measured with passed=false, choose needs_implementation or needs_richer_proof instead of ready_to_ship.',
|
|
1123
1154
|
'For data/audio/log/metrics/custom modes, judge the structured evidence bundle and proof_evidence_sample directly; screenshots are optional supporting context.',
|
|
1124
1155
|
'The summary must name the concrete change, the target route/UI, what changed in after evidence, and why the stop condition is satisfied.',
|
|
1125
1156
|
'Only set escalation_target=human when you conclude the workflow has hit a real wall or is not converging.',
|
|
@@ -1374,6 +1405,9 @@ s['evidence_bundle'] = evidence_bundle
|
|
|
1374
1405
|
visual_delta = ((evidence_bundle.get('after') or {}).get('visual_delta') or {})
|
|
1375
1406
|
if visual_delta.get('status') != 'not_applicable':
|
|
1376
1407
|
summary_lines.append('Visual delta gate: ' + compact_value(visual_delta, limit=700))
|
|
1408
|
+
visual_delta_blocker = visual_delta_blocker_for_mode(s.get('verification_mode'), visual_delta)
|
|
1409
|
+
if visual_delta_blocker:
|
|
1410
|
+
summary_lines.append('Visual delta hard gate: ' + visual_delta_blocker)
|
|
1377
1411
|
|
|
1378
1412
|
proof_evidence_blocker = ''
|
|
1379
1413
|
if proof_evidence_required_for_mode(s.get('verification_mode')):
|
|
@@ -1200,17 +1200,19 @@ def run_verify_requests_supervisor_assessment():
|
|
|
1200
1200
|
artifact_contract = after_verify['proof_assessment_request']['artifact_contract']
|
|
1201
1201
|
assert artifact_contract['required']['baseline_context'] is True
|
|
1202
1202
|
assert artifact_contract['required']['screenshot'] is True
|
|
1203
|
+
assert artifact_contract['required']['visual_delta'] is True
|
|
1203
1204
|
artifact_production = after_verify['proof_assessment_request']['artifact_production']
|
|
1204
1205
|
assert artifact_production['image_output_count'] >= 1
|
|
1205
1206
|
assert artifact_production['proof_evidence_present'] is False
|
|
1206
1207
|
artifact_usage = after_verify['proof_assessment_request']['artifact_usage']
|
|
1207
|
-
assert artifact_usage['missing_required_signals'] == []
|
|
1208
|
+
assert artifact_usage['missing_required_signals'] == ['visual_delta']
|
|
1208
1209
|
assert 'after-capture' in artifact_usage['supervisor_review_signals']
|
|
1209
1210
|
assert 'baseline_context' in artifact_usage['required_signals']
|
|
1210
1211
|
assert 'route_semantics' in artifact_usage['available_signals']
|
|
1212
|
+
assert 'visual_delta.status=unmeasured' in after_verify['proof_assessment_request']['hard_blockers'][0]
|
|
1211
1213
|
assert after_verify['proof_assessment_request']['evidence_bundle']['artifact_contract']['required']['screenshot'] is True
|
|
1212
1214
|
assert after_verify['proof_assessment_request']['evidence_bundle']['artifact_production']['image_output_count'] >= 1
|
|
1213
|
-
assert after_verify['proof_assessment_request']['evidence_bundle']['artifact_usage']['missing_required_signals'] == []
|
|
1215
|
+
assert after_verify['proof_assessment_request']['evidence_bundle']['artifact_usage']['missing_required_signals'] == ['visual_delta']
|
|
1214
1216
|
assert 'capture success is not proof' in '\n'.join(after_verify['proof_assessment_request']['instructions'])
|
|
1215
1217
|
assert after_verify['verify_decision_request']['continue_with_stage'] is None
|
|
1216
1218
|
assert after_verify['verify_results']['baseline']['before']['source'] == 'recon'
|
|
@@ -1573,6 +1575,62 @@ def run_ship_missing_supervisor_gate():
|
|
|
1573
1575
|
shutil.rmtree(tempdir, ignore_errors=True)
|
|
1574
1576
|
|
|
1575
1577
|
|
|
1578
|
+
def run_ship_blocks_unmeasured_visual_delta():
|
|
1579
|
+
tempdir = Path(tempfile.mkdtemp(prefix='riddle-proof-ship-visual-delta-'))
|
|
1580
|
+
state_path = tempdir / 'state.json'
|
|
1581
|
+
try:
|
|
1582
|
+
state = base_state(tempdir, reference='before')
|
|
1583
|
+
state.update({
|
|
1584
|
+
'recon_status': 'ready_for_proof_plan',
|
|
1585
|
+
'author_status': 'ready',
|
|
1586
|
+
'proof_plan_status': 'ready',
|
|
1587
|
+
'implementation_status': 'changes_detected',
|
|
1588
|
+
'verification_mode': 'visual',
|
|
1589
|
+
'verify_status': 'evidence_captured',
|
|
1590
|
+
'before_cdn': 'https://cdn.example.com/before.png',
|
|
1591
|
+
'after_cdn': 'https://cdn.example.com/after.png',
|
|
1592
|
+
'proof_assessment': {
|
|
1593
|
+
'decision': 'ready_to_ship',
|
|
1594
|
+
'summary': 'The screenshots look good.',
|
|
1595
|
+
'source': 'supervising_agent',
|
|
1596
|
+
},
|
|
1597
|
+
'proof_assessment_source': 'supervising_agent',
|
|
1598
|
+
'evidence_bundle': {
|
|
1599
|
+
'verification_mode': 'visual',
|
|
1600
|
+
'artifact_contract': {
|
|
1601
|
+
'required': {
|
|
1602
|
+
'baseline_context': True,
|
|
1603
|
+
'route_semantics': True,
|
|
1604
|
+
'screenshot': True,
|
|
1605
|
+
'visual_delta': True,
|
|
1606
|
+
},
|
|
1607
|
+
},
|
|
1608
|
+
'after': {
|
|
1609
|
+
'screenshot_url': 'https://cdn.example.com/after.png',
|
|
1610
|
+
'observation': {'valid': True, 'reason': 'ok'},
|
|
1611
|
+
'visual_delta': {
|
|
1612
|
+
'status': 'unmeasured',
|
|
1613
|
+
'passed': None,
|
|
1614
|
+
'reason': 'No measured before/after visual delta was found in proof evidence.',
|
|
1615
|
+
},
|
|
1616
|
+
},
|
|
1617
|
+
},
|
|
1618
|
+
})
|
|
1619
|
+
write_state(state_path, state)
|
|
1620
|
+
os.environ['RIDDLE_PROOF_STATE_FILE'] = str(state_path)
|
|
1621
|
+
|
|
1622
|
+
try:
|
|
1623
|
+
load_module('ship_blocks_unmeasured_visual_delta', SHIP_PATH)
|
|
1624
|
+
except SystemExit as exc:
|
|
1625
|
+
message = str(exc)
|
|
1626
|
+
assert 'visual_delta.status=unmeasured' in message, message
|
|
1627
|
+
assert 'blocks ready_to_ship' in message, message
|
|
1628
|
+
return {'ok': True, 'error': message}
|
|
1629
|
+
raise AssertionError('ship should have failed when visual delta was unmeasured')
|
|
1630
|
+
finally:
|
|
1631
|
+
shutil.rmtree(tempdir, ignore_errors=True)
|
|
1632
|
+
|
|
1633
|
+
|
|
1576
1634
|
def run_ship_accepts_structured_after_evidence():
|
|
1577
1635
|
tempdir = Path(tempfile.mkdtemp(prefix='riddle-proof-ship-structured-'))
|
|
1578
1636
|
state_path = tempdir / 'state.json'
|
|
@@ -1829,6 +1887,7 @@ if __name__ == '__main__':
|
|
|
1829
1887
|
'verify_capture_retry': run_verify_capture_retry(),
|
|
1830
1888
|
'missing_baseline_guard': run_verify_missing_baseline(),
|
|
1831
1889
|
'ship_supervisor_gate': run_ship_missing_supervisor_gate(),
|
|
1890
|
+
'ship_blocks_unmeasured_visual_delta': run_ship_blocks_unmeasured_visual_delta(),
|
|
1832
1891
|
'ship_structured_after_evidence': run_ship_accepts_structured_after_evidence(),
|
|
1833
1892
|
'ship_discord_thread_target': run_ship_discord_thread_target(),
|
|
1834
1893
|
'ship_filters_tool_noise_when_staging': run_ship_filters_tool_noise_when_staging(),
|