ProjectTriage/verify.py at main · AshtonVaughan/ProjectTriage · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
#!/usr/bin/env python3
"""Project Triage - Full logic path verification.

Simulates an entire hunt cycle without an LLM. Tests every code path
the agent takes during Phase A, Phase B, and the main loop setup.
Catches type errors, missing attributes, broken imports, and logic bugs.

Usage: PYTHONPATH=. python3 verify.py
   or: python verify.py  (from project root)
"""

import importlib
import json
import os
import sys
import tempfile
import shutil
import traceback
from pathlib import Path
from typing import Any
from unittest.mock import MagicMock

PASS = "\033[92mPASS\033[0m"
FAIL = "\033[91mFAIL\033[0m"
WARN = "\033[93mWARN\033[0m"

errors: list[str] = []
warnings: list[str] = []
passed = 0


def check(name: str, fn):
    global passed
    try:
        fn()
        print(f"  [{PASS}] {name}")
        passed += 1
    except Exception as e:
        tb = traceback.format_exc().strip().split("\n")
        location = tb[-2].strip() if len(tb) >= 2 else ""
        msg = f"{name}: {e} ({location})"
        errors.append(msg)
        print(f"  [{FAIL}] {msg}")


def warn_check(name: str, fn):
    global passed
    try:
        fn()
        print(f"  [{PASS}] {name}")
        passed += 1
    except Exception as e:
        warnings.append(f"{name}: {e}")
        print(f"  [{WARN}] {name}: {e}")


# ── Test data ───────────────────────────────────────────────────────────

TEST_URL = "https://example.com"
TEST_TARGET = "example.com"
TEST_TECH = {
    "framework": "nextjs", "cdn": "cloudflare", "waf": "cloudflare",
    "auth": "jwt", "server": "nginx", "api_style": "rest",
    "cloud_provider": "aws", "auth_type": "oauth",
}
TEST_ENDPOINTS = [
    {"url": "https://example.com/api/users", "method": "GET"},
    {"url": "https://example.com/api/orders/123", "method": "GET"},
    {"url": "https://example.com/api/payments", "method": "POST"},
    {"url": "https://example.com/login", "method": "POST"},
    {"url": "https://example.com/graphql", "method": "POST"},
    {"url": "https://example.com/api/upload", "method": "POST"},
    {"url": "https://example.com/api/admin/users", "method": "GET"},
    {"url": "https://example.com/api/v1/search?q=test", "method": "GET"},
]
TEST_JS = """
var socket = new WebSocket('wss://example.com/ws');
socket.onmessage = function(e) { console.log(e.data); };
window.addEventListener('message', function(e) { eval(e.data); });
fetch('/api/internal/config').then(r => r.json());
const API_KEY = 'sk_live_abc123';
"""
TEST_HTML = """
<html><head><title>Example App</title></head>
<body>
<form action="/login" method="POST">
  <input name="username" type="text">
  <input name="password" type="password">
  <input name="csrf_token" type="hidden" value="abc123">
</form>
<a href="/api/docs">API Docs</a>
<script src="/static/app.js"></script>
</body></html>
"""
TEST_HEADERS = {
    "server": "nginx/1.21.0",
    "x-powered-by": "Express",
    "cf-ray": "abc123",
    "content-type": "text/html",
}
TEST_FINDINGS = [
    {"type": "ssrf", "endpoint": "/api/proxy", "severity": "high",
     "title": "SSRF in proxy endpoint", "evidence": "169.254.169.254 accessible"},
    {"type": "xss", "endpoint": "/search", "severity": "medium",
     "title": "Reflected XSS in search", "evidence": "<script>alert(1)</script> reflected"},
]

# Create temp dir for tests
TMPDIR = tempfile.mkdtemp(prefix="triage_verify_")

# ════════════════════════════════════════════════════════════════════════
print("=" * 70)
print("  PROJECT TRIAGE - FULL LOGIC PATH VERIFICATION")
print("=" * 70)

# ── 1. Module imports ───────────────────────────────────────────────────

print("\n=== 1. MODULE IMPORTS (all 110 files) ===")

MODULES = {
    "core": ["agent", "config", "context", "cost_tracker", "orchestrator",
             "parallel", "pentest_tree", "planner", "prompts", "provider",
             "repetition", "scope", "session", "session_manager", "tool_registry"],
    "brain": ["agot_reasoner", "arch_analyzer", "assumption_engine",
              "chain_analyzer", "chain_engine", "client_analyzer",
              "confusion_engine", "coverage_asymmetry", "curriculum",
              "data_manager", "dom_analyzer", "domain_knowledge",
              "edge_analyzer", "escalation_router", "idor_engine",
              "intent_model", "lats_explorer", "mcts_explorer",
              "perceptor", "procedural_memory", "self_reflect",
              "state_machine", "tech_fingerprint", "websocket_tester",
              "workflow_tester", "world_model", "scale_model"],
    "intel": ["callback_server", "campaign_manager", "differential_engine",
              "fuzzer", "h2_desync", "hackerone", "infra_scanner",
              "interactsh_client", "js_analyzer", "mcp_tester",
              "monitor_mode", "nuclei_scan", "osint_engine",
              "program_intel", "source_analyzer", "source_intel",
              "supply_chain"],
    "models": ["attack_graph", "auth_context", "auth_manager", "cvss",
               "disclosures", "evidence", "hypothesis", "knowledge",
               "memory", "models_db", "patterns", "profiles",
               "target_model"],
    "ui": ["live_display", "report", "report_generator", "tui"],
    "utils": ["db", "evidence_collector", "output_summarizer", "quality_gate",
              "response_classifier", "sanitizer", "utils", "validator", "wordlists"],
    "tools": ["analyzer", "browser", "cache_poison", "cloud_meta",
              "cors", "crawler", "crlf", "desync", "discovery",
              "dns_rebind", "exploit", "fetch_page", "fuzzer_tool",
              "graphql", "jwt", "llm_attacks", "oauth", "prompt_inject",
              "proto_pollution", "race", "recon", "register_discovery",
              "register_new", "saml", "scanner", "ssti",
              "subdomain_takeover", "web_search", "xss", "auth_tools"],
}

for pkg, modules in MODULES.items():
    for mod in modules:
        full = f"{pkg}.{mod}"
        check(full, lambda f=full: importlib.import_module(f))


# ── 2. Score normalization ──────────────────────────────────────────────

print("\n=== 2. HYPOTHESIS SCORE NORMALIZATION ===")

def _test_score_norm():
    from core.agent import _to_int_score
    tests = [
        (8, 8), (0, 1), (15, 10), (-1, 1),  # int
        (0.75, 7), (0.0, 1), (1.0, 10), (0.5, 5),  # float 0-1
        (5.5, 5),  # float >1
        ("critical", 10), ("high", 8), ("medium", 6), ("low", 4), ("none", 1),  # str
        ("HIGH", 8), ("Critical", 10),  # case insensitive
        ("unknown", 5),  # unknown string -> default
        (None, 5), ([], 5), ({}, 5),  # weird types -> default
    ]
    for val, expected in tests:
        result = _to_int_score(val)
        assert result == expected, f"_to_int_score({val!r}) = {result}, expected {expected}"

check("_to_int_score: all input types and edge cases", _test_score_norm)


# ── 3. Brain modules - instantiation + full method calls ────────────────

print("\n=== 3. BRAIN MODULE LOGIC PATHS ===")

# 3a. ConfusionEngine
def _test_confusion_full():
    from brain.confusion_engine import ConfusionEngine
    ce = ConfusionEngine()
    # identify_component_stack
    comps = ce.identify_component_stack(TEST_HEADERS, TEST_URL, TEST_TECH)
    assert isinstance(comps, list), f"Expected list, got {type(comps)}"
    # detect_confusion_opportunities
    vectors = ce.detect_confusion_opportunities(comps, TEST_URL)
    assert isinstance(vectors, list)
    # generate_confusion_payloads for each vector
    for v in vectors[:3]:
        payloads = ce.generate_confusion_payloads(v)
        assert isinstance(payloads, list)
    # generate_confusion_hypotheses - the main output
    hyps = ce.generate_confusion_hypotheses(TEST_URL, TEST_TECH, TEST_ENDPOINTS)
    assert isinstance(hyps, list)
    for h in hyps:
        assert isinstance(h.get("novelty"), (int, float)), f"novelty not numeric: {h.get('novelty')!r}"
        assert isinstance(h.get("exploitability"), (int, float)), f"exploitability not numeric: {h.get('exploitability')!r}"
        assert isinstance(h.get("impact"), (int, float)), f"impact not numeric: {h.get('impact')!r}"
        assert h.get("effort") is not None, "effort is None"

check("ConfusionEngine: full pipeline (stack->vectors->payloads->hypotheses)", _test_confusion_full)

# 3b. IDOREngine
def _test_idor_full():
    from brain.idor_engine import IDOREngine
    ie = IDOREngine()
    idor_tests = ie.generate_idor_tests(TEST_ENDPOINTS, {"roles": ["user", "admin"], "tokens": {}})
    assert isinstance(idor_tests, list)
    bola_tests = ie.generate_bola_tests(TEST_ENDPOINTS, TEST_TECH)
    assert isinstance(bola_tests, list)
    all_tests = idor_tests + bola_tests
    hyps = ie.idor_to_hypotheses(all_tests)
    assert isinstance(hyps, list)
    for h in hyps:
        assert "endpoint" in h, f"Missing endpoint in hypothesis: {h}"
        assert "technique" in h, f"Missing technique: {h}"

check("IDOREngine: generate_idor + generate_bola + to_hypotheses", _test_idor_full)

# 3c. ClientAnalyzer
def _test_client_full():
    from brain.client_analyzer import ClientAnalyzer
    ca = ClientAnalyzer()
    pm_findings = ca.analyze_postmessage_surface(TEST_JS, TEST_URL)
    assert isinstance(pm_findings, list)
    ws_eps = ca.extract_ws_endpoints_from_js(TEST_JS, TEST_URL)
    assert isinstance(ws_eps, list)
    hyps = ca.generate_client_hypotheses(TEST_URL, TEST_JS, TEST_TECH)
    assert isinstance(hyps, list)

check("ClientAnalyzer: postmessage + ws_extract + hypotheses", _test_client_full)

# 3d. WebSocketTester - verify no phantom endpoints
def _test_ws_no_phantom():
    from brain.websocket_tester import WebSocketTester
    wst = WebSocketTester()
    # No JS evidence = no endpoints
    eps_no_evidence = wst.discover_ws_endpoints(TEST_URL)
    assert len(eps_no_evidence) == 0, f"Expected 0 endpoints without evidence, got {len(eps_no_evidence)}"
    # With JS evidence = endpoints appear
    eps_with_js = wst.discover_ws_endpoints(TEST_URL, js_content=TEST_JS)
    assert len(eps_with_js) > 0, "Expected endpoints with WebSocket JS evidence"

check("WebSocketTester: no phantom endpoints without evidence", _test_ws_no_phantom)

# 3e. AssumptionEngine
def _test_assumption_full():
    from brain.assumption_engine import AssumptionEngine
    ae = AssumptionEngine()
    assumptions = ae.generate_assumptions(TEST_URL, "POST", ["user_id", "amount"], TEST_TECH)
    assert isinstance(assumptions, list)
    if assumptions:
        hyps = ae.assumptions_to_hypotheses(assumptions[:3])
        assert isinstance(hyps, list)
        for h in hyps:
            assert "endpoint" in h or "description" in h

check("AssumptionEngine: generate + to_hypotheses", _test_assumption_full)

# 3f. IntentModel
def _test_intent_full():
    from brain.intent_model import IntentModel
    im = IntentModel()
    violations = im.generate_violation_tests(TEST_URL, "POST", ["amount", "status"], TEST_TECH)
    assert isinstance(violations, list)
    if violations:
        hyps = im.violations_to_hypotheses(violations[:3])
        assert isinstance(hyps, list)

check("IntentModel: generate_violations + to_hypotheses", _test_intent_full)

# 3g. EdgeAnalyzer
def _test_edge_full():
    from brain.edge_analyzer import EdgeAnalyzer
    ea = EdgeAnalyzer()
    components = ea.identify_components(TEST_URL, TEST_HEADERS, TEST_TECH)
    assert isinstance(components, list)
    if len(components) >= 2:
        edges = ea.generate_edge_tests(components, TEST_URL)
        assert isinstance(edges, list)
        for edge in edges[:3]:
            assert hasattr(edge, "data_type"), f"Edge missing data_type: {edge}"
            assert hasattr(edge, "upstream"), f"Edge missing upstream: {edge}"

check("EdgeAnalyzer: identify_components + generate_edge_tests", _test_edge_full)

# 3h. DomainKnowledge
def _test_domain_full():
    from brain.domain_knowledge import DomainKnowledge
    dk = DomainKnowledge()
    domain = dk.detect_domain(TEST_URL, [e["url"] for e in TEST_ENDPOINTS], TEST_TECH)
    assert isinstance(domain, str)
    patterns = dk.get_patterns(domain)
    assert isinstance(patterns, list)
    if patterns:
        hyps = dk.patterns_to_hypotheses(patterns[:3], TEST_URL)
        assert isinstance(hyps, list)
        for h in hyps:
            # Verify types won't break _to_int_score
            impact = h.get("impact", "")
            assert impact is not None, f"impact is None in hypothesis: {h}"

check("DomainKnowledge: detect_domain + get_patterns + to_hypotheses", _test_domain_full)

# 3i. ArchAnalyzer
def _test_arch_full():
    from brain.arch_analyzer import ArchAnalyzer
    aa = ArchAnalyzer()
    # detect_patterns - exact agent code path uses (url, tech_stack, headers)
    patterns = aa.detect_patterns(TEST_URL, TEST_TECH, TEST_HEADERS)
    assert isinstance(patterns, list)
    # Agent accesses pattern.name, pattern.description, pattern.severity directly
    for p in patterns[:5]:
        assert hasattr(p, "name"), f"Pattern missing 'name': {type(p)}"
        assert hasattr(p, "description"), f"Pattern missing 'description'"
        assert hasattr(p, "severity"), f"Pattern missing 'severity'"
        assert isinstance(p.name, str), f"name not str: {type(p.name)}"
        assert p.severity in ("critical", "high", "medium", "low", "info"), f"unexpected severity: {p.severity}"

check("ArchAnalyzer: detect_patterns + pattern attributes (exact agent path)", _test_arch_full)

# 3j. ChainAnalyzer + ChainEngine
def _test_chain_full():
    from brain.chain_analyzer import ChainAnalyzer
    from brain.chain_engine import ChainEngine
    ca = ChainAnalyzer()
    # Test with findings
    chains = ca.analyze(TEST_FINDINGS)
    assert isinstance(chains, list)
    if chains:
        hyps = ca.get_chain_hypotheses(chains)
        assert isinstance(hyps, list)
    # ChainEngine
    ce = ChainEngine()
    chain_hyps = ce.generate_chain_hypotheses(TEST_FINDINGS, TEST_URL)
    assert isinstance(chain_hyps, list)

check("ChainAnalyzer + ChainEngine: analyze + hypotheses", _test_chain_full)

# 3k. ScaleModel
def _test_scale_full():
    from brain.scale_model import ScaleModel
    sm = ScaleModel()
    # estimate_scale
    scale = sm.estimate_scale(TEST_TECH, subdomain_count=50, endpoint_count=200,
                              program_info={"bounty_max": 15000})
    assert scale.scale_tier in ("startup", "mid_market", "enterprise", "mega")
    assert 1 <= scale.attack_surface_score <= 100
    # get_scale_hypotheses
    hyps = sm.get_scale_hypotheses(TEST_TECH, 50, 200)
    assert isinstance(hyps, list)
    for h in hyps:
        from core.agent import _to_int_score
        # Verify all fields can be normalized without error
        _to_int_score(h.get("novelty", 5))
        _to_int_score(h.get("exploitability", 5))
        _to_int_score(h.get("impact", 5))
        _to_int_score(h.get("effort", 5))
    # environment hypotheses
    env_hyps = sm.get_environment_hypotheses("example.com")
    assert isinstance(env_hyps, list)
    assert len(env_hyps) > 0, "Expected environment hypotheses"
    # role escalation
    matrix = sm.get_role_escalation_matrix(["user", "admin", "moderator"])
    assert isinstance(matrix, list)
    assert len(matrix) >= 6, f"Expected 6+ escalation pairs for 3 roles, got {len(matrix)}"

check("ScaleModel: estimate + hypotheses + environments + role_matrix", _test_scale_full)

# 3l. CoverageAsymmetry
def _test_coverage_full():
    from brain.coverage_asymmetry import CoverageAsymmetryDetector
    cd = CoverageAsymmetryDetector()
    for ep in TEST_ENDPOINTS:
        assessment = cd.assess_surface(ep["url"])
        assert hasattr(assessment, "estimated_coverage")
        assert hasattr(assessment, "priority_boost")
        assert isinstance(assessment.priority_boost, (int, float))

check("CoverageAsymmetry: assess_surface on all test endpoints", _test_coverage_full)

# 3m. DOMAnalyzer
def _test_dom_full():
    from brain.dom_analyzer import DOMAnalyzer
    da = DOMAnalyzer()
    hyps = da.generate_hypotheses(TEST_URL, TEST_TECH)
    assert isinstance(hyps, list)

check("DOMAnalyzer: generate_hypotheses", _test_dom_full)

# 3n. ProceduralMemory
def _test_procedural_full():
    from brain.procedural_memory import ProceduralMemory, AttackTrace
    pm_dir = Path(TMPDIR) / "pm_test"
    pm_dir.mkdir(exist_ok=True)
    pm = ProceduralMemory(pm_dir)
    # Find applicable skills
    skills = pm.find_applicable_skills({"framework": "express", "auth": "jwt"}, TEST_URL)
    assert isinstance(skills, list)
    # Get hypotheses from skills
    hyps = pm.get_skill_hypotheses(skills, TEST_TARGET)
    assert isinstance(hyps, list)
    for h in hyps:
        assert "endpoint" in h
        assert "technique" in h
    # Record a trace
    trace = AttackTrace(
        target=TEST_TARGET, technique="sqli", finding_severity="high",
        tech_stack=TEST_TECH, timestamp="2026-01-01T00:00:00",
        steps=[{"tool": "sqlmap", "inputs": {"url": TEST_URL}, "output_summary": "injectable", "success": True}],
    )
    pm.record_trace(trace)
    # Compile skill
    skill = pm.compile_skill(trace)
    assert skill is not None
    # Update stats
    pm.update_skill_stats(skill.skill_id, True)
    pm.update_skill_stats(skill.skill_id, False)
    # Summary
    summary = pm.get_skill_summary()
    assert isinstance(summary, str)
    # Close DB connection to prevent file locks
    if hasattr(pm, '_conn') and pm._conn:
        try:
            pm._conn.close()
        except Exception:
            pass

warn_check("ProceduralMemory: find_skills + hypotheses + record + compile + stats", _test_procedural_full)

# 3o. Curriculum
def _test_curriculum_full():
    from brain.curriculum import CurriculumManager
    cm = CurriculumManager(Path(TMPDIR) / "cur_test")
    # Assess difficulty
    profile = cm.assess_target_difficulty(TEST_TECH, "cloudflare", "oauth")
    assert hasattr(profile, "level")
    assert 1 <= profile.level <= 10
    assert hasattr(profile, "name")
    # Get hypotheses
    hyps = cm.get_curriculum_hypotheses(profile.level, TEST_TECH)
    assert isinstance(hyps, list)
    # Update mastery
    cm.update_mastery("sqli_union", True, "high")
    cm.update_mastery("sqli_union", False, "")
    # Report
    report = cm.get_mastery_report()
    assert isinstance(report, str)

check("Curriculum: assess + hypotheses + mastery_update + report", _test_curriculum_full)

# 3p. EscalationRouter
def _test_escalation_full():
    from brain.escalation_router import EscalationRouter
    mock_provider = MagicMock()
    mock_config = MagicMock()
    mock_config.data_dir = Path(TMPDIR)
    er = EscalationRouter(mock_provider, mock_config)
    assert er.frontier_available == False, "Should be False without API key"
    # should_escalate
    result = er.should_escalate("hypothesis_generation", 0.2, {})
    assert isinstance(result, bool)
    # Stats
    stats = er.get_escalation_stats()
    assert isinstance(stats, dict)

check("EscalationRouter: frontier_available + should_escalate + stats", _test_escalation_full)

# 3q. DataManager
def _test_data_manager_full():
    from brain.data_manager import DataManager
    dm = DataManager(Path("data"))
    # Routes for tech
    routes = dm.get_routes_for_tech(TEST_TECH)
    assert isinstance(routes, list)
    assert len(routes) > 0, "Expected routes for nextjs tech stack"
    # Params for tech
    params = dm.get_params_for_tech(TEST_TECH)
    assert isinstance(params, list)
    # Sensitive paths
    paths = dm.get_sensitive_paths()
    assert isinstance(paths, list)
    assert len(paths) > 50, f"Expected 50+ sensitive paths, got {len(paths)}"
    # API patterns
    api = dm.get_api_patterns("rest")
    assert isinstance(api, list)
    # WAF bypass payloads
    waf = dm.get_waf_bypass_payloads("cloudflare", "xss")
    assert isinstance(waf, list)
    # Backup extensions
    backups = dm.get_backup_extensions()
    assert isinstance(backups, list)

check("DataManager: routes + params + paths + api + waf_bypass + backups", _test_data_manager_full)


# ── 4. Intel modules ───────────────────────────────────────────────────

print("\n=== 4. INTEL MODULE LOGIC PATHS ===")

def _test_h1_importer_full():
    from intel.hackerone import HackerOneImporter, ProgramProfile, BountyTable, ScopeAsset
    h1 = HackerOneImporter(Path(TMPDIR) / "h1_test")
    # Parse input
    for inp, expected_handle, expected_platform in [
        ("shopify", "shopify", "hackerone"),
        ("https://hackerone.com/shopify", "shopify", "hackerone"),
        ("https://hackerone.com/shopify/policy_scopes", "shopify", "hackerone"),
        ("https://bugcrowd.com/tesla", "tesla", "bugcrowd"),
    ]:
        h, p = h1._parse_input(inp)
        assert h == expected_handle, f"parse_input({inp!r}): expected handle={expected_handle}, got {h}"
        assert p == expected_platform, f"parse_input({inp!r}): expected platform={expected_platform}, got {p}"
    # BountyTable
    bt = BountyTable(critical_min=5000, critical_max=20000, high_min=1000, high_max=5000)
    assert bt.severity_value("critical") == 20000
    assert bt.severity_value("high") == 5000
    d = bt.to_dict()
    assert d["critical_max"] == 20000
    # ScopeAsset
    sa = ScopeAsset(asset_type="URL", identifier="*.example.com", eligible_for_bounty=True,
                    eligible_for_submission=True, instruction="test", max_severity="critical",
                    created_at="2026-01-01", confidentiality_requirement="high")
    assert sa.matches_url("https://sub.example.com/path")
    assert not sa.matches_url("https://other.com")
    # ProgramProfile
    pp = ProgramProfile(handle="test", name="Test", url="https://hackerone.com/test",
                        platform="hackerone", state="open", offers_bounties=True,
                        bounty_table=bt, in_scope=[sa])
    assert pp.is_in_scope("https://api.example.com/users")
    assert not pp.is_in_scope("https://notinscope.com")
    # Serialize round-trip
    d = pp.to_dict()
    pp2 = ProgramProfile.from_dict(d)
    assert pp2.handle == "test"
    assert len(pp2.in_scope) == 1
    # Save + load
    h1.save_program(pp)
    loaded = h1.load_program("test")
    assert loaded is not None
    assert loaded.handle == "test"
    # Generate scope context
    ctx = h1.generate_scope_context(pp)
    assert "In-Scope" in ctx
    assert "*.example.com" in ctx
    # List programs
    programs = h1.list_saved_programs()
    assert len(programs) >= 1

check("HackerOneImporter: parse + BountyTable + ScopeAsset + ProgramProfile + save/load + context", _test_h1_importer_full)

def _test_program_intel():
    from intel.program_intel import ProgramIntelligence
    pi = ProgramIntelligence()
    handle, platform = pi.extract_handle_from_url(TEST_TARGET)
    assert isinstance(handle, str)
    assert isinstance(platform, str)

check("ProgramIntelligence: extract_handle", _test_program_intel)

def _test_campaign_mgr():
    from intel.campaign_manager import CampaignManager
    cm = CampaignManager(Path(TMPDIR) / "campaign_test")
    campaign = cm.create_campaign(TEST_TARGET)
    assert campaign is not None
    assert campaign.target == TEST_TARGET
    found = cm.find_campaign_for_target(TEST_TARGET)
    assert found is not None
    cm.start_session(campaign)

check("CampaignManager: create + find + start_session", _test_campaign_mgr)


# ── 5. Models ──────────────────────────────────────────────────────────

print("\n=== 5. MODEL LOGIC PATHS ===")

def _test_auth_context_full():
    from models.auth_context import AuthContext
    ac = AuthContext()
    s1 = ac.add_session("user_a", "user", cookies={"session": "abc"}, jwt_token="eyJ...")
    s2 = ac.add_session("admin", "admin", headers={"Authorization": "Bearer xyz"})
    s3 = ac.add_session("unauth", "unauthenticated")
    # Test the extraction pattern used in agent.py
    roles = [s.role for s in ac.sessions.values()]
    assert "user" in roles
    assert "admin" in roles
    tokens = {n: bool(s.jwt_token) for n, s in ac.sessions.items()}
    assert tokens["user_a"] == True
    assert tokens["admin"] == False
    # Hypothesis generation
    hyps = ac.get_all_test_hypotheses()
    assert isinstance(hyps, list)

check("AuthContext: add_session + role extraction + token extraction + hypotheses", _test_auth_context_full)

def _test_hypothesis_engine():
    from models.hypothesis import HypothesisEngine, Hypothesis
    from utils.db import Database
    db = Database(Path(TMPDIR) / "hyp_test.db")
    he = HypothesisEngine(db, TEST_TARGET)
    h = he.create(
        endpoint=TEST_URL, technique="sqli",
        description="SQL injection in search",
        novelty=7, exploitability=8, impact=9, effort=3,
    )
    assert h is not None
    assert isinstance(h, Hypothesis)
    assert h.total_score > 0
    db.close()

check("HypothesisEngine: create hypothesis with scoring", _test_hypothesis_engine)

def _test_attack_graph():
    from models.attack_graph import AttackGraph
    from utils.db import Database
    db = Database(Path(TMPDIR) / "ag_test.db")
    ag = AttackGraph(db, TEST_TARGET, max_steps=10)
    assert not ag.is_complete
    db.close()

check("AttackGraph: create + is_complete", _test_attack_graph)

def _test_world_model():
    from brain.world_model import WorldModel
    wm = WorldModel(TEST_TARGET, Path(TMPDIR) / "wm_test")
    wm.add_host(TEST_TARGET, {"port": 443, "service": "https"})
    wm.set_tech("framework", "nextjs")
    wm.set_tech("cdn", "cloudflare")
    tech = wm.tech_stack
    assert tech.get("framework") == "nextjs"
    assert tech.get("cdn") == "cloudflare"
    # Add finding
    wm.add_finding(
        id="f1", title="XSS in search", severity="medium",
        description="Reflected XSS", endpoint="/search",
        technique="xss", step_found=1,
        chain_potential=["xss", "cache_poison"],
    )
    findings = wm.get_findings_for_chain_analysis()
    assert isinstance(findings, list)

check("WorldModel: add_host + set_tech + add_finding + chain_analysis", _test_world_model)

def _test_target_model():
    from models.target_model import TargetModel
    tm = TargetModel(TEST_TARGET, Path(TMPDIR) / "tm_test")
    assert tm.safe_name is not None
    assert isinstance(tm.is_stale, bool)

check("TargetModel: create + safe_name + is_stale", _test_target_model)

def _test_quality_gate():
    from utils.quality_gate import QualityGate
    qg = QualityGate()
    finding = {
        "title": "XSS in search",
        "endpoint": "/search?q=<script>",
        "severity": "medium",
        "evidence": "Script tag reflected in response",
        "technique": "xss",
        "confidence": 0.8,
    }
    score = qg.score_finding(finding)
    assert score is not None, "score_finding returned None"
    surface = qg.should_surface(score)
    assert isinstance(surface, bool)
    # Test noise detection
    is_noise, reason = qg.is_noise("xss", "XSS in search parameter")
    assert isinstance(is_noise, bool)

check("QualityGate: score_finding + should_surface + is_noise", _test_quality_gate)


# ── 6. Tool registrations ──────────────────────────────────────────────

print("\n=== 6. TOOL REGISTRATIONS ===")

def _test_all_tool_regs():
    try:
        from core.config import Config
        config = Config.from_env()
    except RuntimeError:
        return  # No tools on PATH

    from main import build_registry
    registry = build_registry(config)
    assert len(registry.tools) > 30, f"Expected 30+ tools, got {len(registry.tools)}"
    # Verify every tool has required attributes
    for name, tool in registry.tools.items():
        assert tool.name, f"Tool has empty name"
        assert tool.description, f"Tool {name} has empty description"
        assert tool.parameters is not None, f"Tool {name} has None parameters"
        assert callable(tool.execute), f"Tool {name}.execute is not callable"
        assert isinstance(tool.phase_tags, list), f"Tool {name}.phase_tags is not a list"
        # Verify to_description doesn't crash
        desc = tool.to_description()
        assert isinstance(desc, str)
        # Verify to_embed_text doesn't crash
        embed = tool.to_embed_text()
        assert isinstance(embed, str)

warn_check("build_registry: all tools valid", _test_all_tool_regs)


# ── 7. Agent import chain ──────────────────────────────────────────────

print("\n=== 7. AGENT IMPORT CHAIN ===")

check("from core.agent import Agent, _to_int_score",
      lambda: __import__("core.agent", fromlist=["Agent", "_to_int_score"]))

check("from main import build_registry, main, _run_hunt",
      lambda: __import__("main", fromlist=["build_registry", "main", "_run_hunt"]))

check("from ui.tui import main_menu",
      lambda: __import__("ui.tui", fromlist=["main_menu"]))


# ── 8. Cross-module integration ────────────────────────────────────────

print("\n=== 8. CROSS-MODULE INTEGRATION ===")

def _test_confusion_through_agent_scorer():
    """Test the exact code path agent.py uses for confusion hypotheses."""
    from brain.confusion_engine import ConfusionEngine
    from core.agent import _to_int_score
    ce = ConfusionEngine()
    hyps = ce.generate_confusion_hypotheses(TEST_URL, TEST_TECH, TEST_ENDPOINTS)
    for h in hyps:
        n = _to_int_score(h.get("novelty", 9))
        e = _to_int_score(h.get("exploitability", 8))
        i = _to_int_score(h.get("impact", 9))
        eff = _to_int_score(h.get("effort", 4))
        assert 1 <= n <= 10, f"novelty out of range: {n} from {h.get('novelty')!r}"
        assert 1 <= e <= 10, f"exploitability out of range: {e}"
        assert 1 <= i <= 10, f"impact out of range: {i}"
        assert 1 <= eff <= 10, f"effort out of range: {eff} from {h.get('effort')!r}"

check("Confusion -> _to_int_score integration (exact agent code path)", _test_confusion_through_agent_scorer)

def _test_scale_through_agent_scorer():
    """Test the exact code path agent.py uses for scale hypotheses."""
    from brain.scale_model import ScaleModel
    from core.agent import _to_int_score
    sm = ScaleModel()
    hyps = sm.get_scale_hypotheses(TEST_TECH, 10, 50)
    for h in hyps:
        n = _to_int_score(h.get("novelty", 6))
        e = _to_int_score(h.get("exploitability", 5))
        i = _to_int_score(h.get("impact", 7))
        eff = _to_int_score(h.get("effort", 3))
        assert 1 <= n <= 10
        assert 1 <= e <= 10
        assert 1 <= i <= 10
        assert 1 <= eff <= 10

check("Scale -> _to_int_score integration (exact agent code path)", _test_scale_through_agent_scorer)

def _test_domain_knowledge_through_agent_scorer():
    """Test the exact code path agent.py uses for domain knowledge hypotheses."""
    from brain.domain_knowledge import DomainKnowledge
    from core.agent import _to_int_score
    dk = DomainKnowledge()
    domain = dk.detect_domain(TEST_URL, [], TEST_TECH)
    patterns = dk.get_patterns(domain)
    if patterns:
        hyps = dk.patterns_to_hypotheses(patterns[:5], TEST_URL)
        for h in hyps:
            i = _to_int_score(h.get("impact", h.get("severity", 8)))
            assert 1 <= i <= 10, f"impact out of range: {i} from {h.get('impact')!r} / {h.get('severity')!r}"

check("DomainKnowledge -> _to_int_score integration (exact agent code path)", _test_domain_knowledge_through_agent_scorer)

def _test_env_hypotheses_through_agent_scorer():
    """Test environment hypotheses have valid scores."""
    from brain.scale_model import ScaleModel
    from core.agent import _to_int_score
    sm = ScaleModel()
    hyps = sm.get_environment_hypotheses("example.com")
    assert len(hyps) > 0
    for h in hyps[:20]:
        for field in ["novelty", "exploitability", "impact", "effort"]:
            val = h.get(field, 5)
            score = _to_int_score(val)
            assert 1 <= score <= 10, f"env hyp {field}={val!r} -> {score} out of range"

check("Environment hypotheses -> _to_int_score (all 93 hypotheses)", _test_env_hypotheses_through_agent_scorer)


# ── 9. Reasoning quality modules ───────────────────────────────────────

print("\n=== 9. REASONING QUALITY MODULES ===")

def _test_repetition_identifier():
    from core.repetition import RepetitionIdentifier
    ri = RepetitionIdentifier()
    # First two calls should pass
    b1, _ = ri.check("curl", {"url": "https://x.com/ws"}, 1)
    ri.record("curl", {"url": "https://x.com/ws"}, 1)
    b2, _ = ri.check("curl", {"url": "https://x.com/ws"}, 2)
    ri.record("curl", {"url": "https://x.com/ws"}, 2)
    # Third should be blocked
    b3, reason = ri.check("curl", {"url": "https://x.com/ws"}, 3)
    assert b3, "Should block after 2 exact repeats"
    assert "BLOCK" in reason.upper()
    # Different inputs should pass
    b4, _ = ri.check("curl", {"url": "https://x.com/api"}, 4)
    assert not b4
    # Get untried tools
    untried = ri.get_untried_tools(["nmap", "curl", "subfinder", "nuclei"])
    assert "nmap" in untried  # nmap hasn't been used
    assert "curl" not in untried  # curl has been used

check("RepetitionIdentifier: block repeats + allow different + untried tools", _test_repetition_identifier)

def _test_pentest_tree():
    from core.pentest_tree import PentestTree
    pt = PentestTree("example.com")
    pt.add_service("example.com", 443, "https", "nginx/1.21")
    pt.add_service("example.com", 80, "http")
    pt.add_subdomain("api.example.com")
    pt.add_subdomain("admin.example.com")
    pt.add_tech("framework", "nextjs")
    pt.add_tech("waf", "cloudflare")
    pt.record_success(1, "nmap", "example.com", "Found 2 open ports")
    pt.record_failure(2, "curl", "wss://example.com/ws", "404 not found")
    pt.record_failure(3, "curl", "wss://example.com/socket", "404 not found")
    pt.block_path("No WebSocket endpoints exist on this target")
    pt.add_finding({"type": "xss", "endpoint": "/search", "severity": "medium"})
    # Test was_tried
    assert pt.was_tried("nmap", "example.com")
    assert pt.was_tried("curl", "wss://example.com/ws")
    assert not pt.was_tried("nuclei", "example.com")
    # Render
    rendered = pt.render(max_chars=3000)
    assert len(rendered) <= 3000
    assert "example.com" in rendered
    assert "nginx" in rendered or "443" in rendered
    # Serialize round-trip
    d = pt.to_dict()
    pt2 = PentestTree.from_dict(d)
    assert pt2.target == "example.com"
    assert len(pt2.services) == 2

check("PentestTree: full lifecycle + render + serialize", _test_pentest_tree)

def _test_response_classifier():
    from utils.response_classifier import ResponseClassifier
    rc = ResponseClassifier()
    # WAF block
    r1 = rc.classify_http(403, {"server": "cloudflare", "cf-ray": "abc"}, "Attention Required! Cloudflare")
    assert r1.category == "waf_block", f"Expected waf_block, got {r1.category}"
    assert r1.waf_vendor == "cloudflare"
    # Rate limit
    r2 = rc.classify_http(429, {}, "Too many requests")
    assert r2.category == "rate_limited"
    # Auth redirect
    r3 = rc.classify_http(302, {"location": "/login"}, "")
    assert r3.category == "auth_redirect"
    # Real content
    r4 = rc.classify_http(200, {"content-type": "text/html"}, "<html><body>Hello World</body></html>")
    assert r4.category == "real_content"
    # Tool error
    r5 = rc.classify_tool_output("nmap", -1, "", "Command timed out after 120s")
    assert r5.category == "tool_error"
    # Empty suspicious
    r6 = rc.classify_tool_output("nuclei", 0, "", "")
    assert r6.category in ("tool_error", "empty_suspicious")
    # Format for agent
    fmt = rc.format_for_agent(r1)
    assert "WAF" in fmt.upper() or "BLOCK" in fmt.upper()

check("ResponseClassifier: WAF + rate_limit + auth_redirect + real + tool_error + format", _test_response_classifier)

def _test_output_summarizer():
    from utils.output_summarizer import OutputSummarizer
    os_mod = OutputSummarizer()
    # nmap
    s1 = os_mod.summarize("nmap", "80/tcp open http nginx/1.21\n443/tcp open ssl/https\n22/tcp filtered ssh")
    assert "80" in s1 or "open" in s1.lower()
    # subfinder
    s2 = os_mod.summarize("subfinder", "api.example.com\nadmin.example.com\nstaging.example.com")
    assert "3" in s2 or "api" in s2
    # empty
    s3 = os_mod.summarize("nuclei", "")
    assert s3  # Should return something, not empty
    # curl
    s4 = os_mod.summarize("curl", "HTTP/1.1 200 OK\nContent-Type: text/html\n\n<html>test</html>")
    assert "200" in s4
    # is_empty_or_blocked
    empty, reason = os_mod.is_empty_or_blocked("nuclei", "")
    assert empty

check("OutputSummarizer: nmap + subfinder + empty + curl + is_empty_or_blocked", _test_output_summarizer)

def _test_session_manager():
    from core.session_manager import SessionManager
    import tempfile
    td = Path(tempfile.mkdtemp())
    sm = SessionManager(td)
    # Add credentials
    cred = sm.add_credential("user_a", "test@test.com", "pass123", role="user")
    assert cred.username == "test@test.com"
    cred2 = sm.add_credential("admin", "admin@test.com", "adminpass", role="admin")
    # List (masked)
    creds = sm.list_credentials()
    assert len(creds) == 2
    # Has multiple users
    assert not sm.has_multiple_users()  # No active sessions yet
    # Auth context
    ctx = sm.get_auth_context_for_agent()
    assert isinstance(ctx, str)
    # Remove
    sm.remove_credential("admin")
    assert len(sm.credentials) == 1

check("SessionManager: add/remove credentials + list + auth context", _test_session_manager)

def _test_constrained_prompt_wiring():
    """Verify the constrained prompt is actually used in _agent_step."""
    import inspect
    from core.agent import Agent
    source = inspect.getsource(Agent._agent_step)
    assert "CONSTRAINED_ACTION_PROMPT" in source, "CONSTRAINED_ACTION_PROMPT not used in _agent_step"
    assert "_build_action_list" in source, "_build_action_list not called in _agent_step"

check("Constrained prompt wired into _agent_step (not just defined)", _test_constrained_prompt_wiring)

def _test_scope_checking():
    """Verify scope checking exists in _execute_tool."""
    import inspect
    from core.agent import Agent
    source = inspect.getsource(Agent._execute_tool)
    assert "_check_scope" in source, "_check_scope not called in _execute_tool"
    assert "OUT OF SCOPE" in source, "Out of scope message not in _execute_tool"

check("Scope checking wired into _execute_tool", _test_scope_checking)

def _test_throttling():
    """Verify adaptive throttling exists in _execute_tool."""
    import inspect
    from core.agent import Agent
    source = inspect.getsource(Agent._execute_tool)
    assert "_throttle_seconds" in source, "Throttling not in _execute_tool"

check("Adaptive throttling wired into _execute_tool", _test_throttling)


# ── Cleanup + Results ──────────────────────────────────────────────────

try:
    shutil.rmtree(TMPDIR, ignore_errors=True)
except Exception:
    pass

print("\n" + "=" * 70)
print(f"  PASSED:   {passed}")
print(f"  FAILED:   {len(errors)}")
print(f"  WARNINGS: {len(warnings)}")
print("=" * 70)

if errors:
    print(f"\n\033[91mFAILURES:\033[0m")
    for e in errors:
        print(f"  - {e}")

if warnings:
    print(f"\n\033[93mWARNINGS (non-critical):\033[0m")
    for w in warnings:
        print(f"  - {w}")

if not errors:
    print(f"\n\033[92mAll {passed} checks passed. Every logic path verified. Safe to hunt.\033[0m")
else:
    print(f"\n\033[91mFix {len(errors)} error(s) before hunting.\033[0m")

sys.exit(1 if errors else 0)