qvm: contact eligibility gates

docs/ASSET_PIPELINE_TODO.md

@@ -766,3 +766,6 @@ Research progress: RE-698 [report](research/MD3_ORIGINAL_MODEL_HIT.md); remainin
 
 
 Research progress: RE-699 [report](research/MD3_GEOMETRY_AUDIT.md); remaining native/content/runtime/platform scope stays open.
+
+
+Research progress: RE-700 [report](research/QVM_CONTACT_GATES.md); remaining native/content/runtime/platform scope stays open.

docs/FINDINGS.md

@@ -1112,3 +1112,6 @@ Bounded qagame paths recover bomb-plant/inactivity correction requests, direct f
 
 
 **RE-699 — Original-asset numerical geometry comparison audit:** Original-asset runs retain measurable source-point differences from conventional ray points and localized fourth-row differences; native/reference validation remains necessary. [Evidence](research/MD3_GEOMETRY_AUDIT.md).
+
+
+**RE-700 — Contact eligibility gates with original-model hit contexts:** 292545 gates client contact on damageable/special mode, nonempty hit context and team/friendly-fire rules before reached damage calculation. [Evidence](research/QVM_CONTACT_GATES.md).

docs/PROVISIONAL_TODO.md

@@ -1098,3 +1098,6 @@ Research progress: RE-698 [report](research/MD3_ORIGINAL_MODEL_HIT.md); remainin
 
 
 Research progress: RE-699 [report](research/MD3_GEOMETRY_AUDIT.md); remaining native/content/runtime/platform scope stays open.
+
+
+Research progress: RE-700 [report](research/QVM_CONTACT_GATES.md); remaining native/content/runtime/platform scope stays open.

docs/RESEARCH_LEDGER.md

@@ -1142,3 +1142,8 @@ Complete144766 now uses original Athena/Orion lower/upper/head/helmet buffers, a
 ## 2026-10-21 — RE-699 Original-asset numerical geometry comparison audit (separate task commit)
 
 Original-asset runs retain measurable source-point differences from conventional ray points and localized fourth-row differences; native/reference validation remains necessary. Four audit checks retain exact input report hashes, independently recompute all point distances, localize matrix differences and reconcile evidence totals. Full suite: 1251 tests pass. Report regenerates byte-identically (38388 bytes). Storage/filter, links and whitespace checks pass. No assets/Rust runtime/shared VM evaluator changed; native/reference/legal/full platform scope remains open.
+
+
+## 2026-10-21 — RE-700 Contact eligibility gates with original-model hit contexts (separate task commit)
+
+292545 gates client contact on damageable/special mode, nonempty hit context and team/friendly-fire rules before reached damage calculation. 288 branch fixtures independently check the complete prefix truth table, mode-call order, original context provenance and unchanged external state. Full suite: 1255 tests pass. Report regenerates byte-identically (245882 bytes). Storage/filter, links and whitespace checks pass. No assets/Rust runtime/shared VM evaluator changed; native/reference/legal/full platform scope remains open.

docs/REVERSE_ENGINEERING_TODO.md

@@ -1141,3 +1141,6 @@ Purpose: turn the supplied installation into an evidence-backed specification of
 
 
 - [x] RE-699 Original-asset numerical geometry comparison audit ([report](research/MD3_GEOMETRY_AUDIT.md)).
+
+
+- [x] RE-700 Contact eligibility gates with original-model hit contexts ([report](research/QVM_CONTACT_GATES.md)).

docs/research/QVM_CONTACT_GATES.md

@@ -0,0 +1,7 @@
+# Contact eligibility gates with original-model hit contexts
+
+RE-700. Reproduce with `python3 scripts/analyze_qvm_contact_gates.py`; check with `python3 scripts/test_qvm_contact_gates.py`. The [machine report](qvm-contact-gates.json) retains source identity and bounded fixtures.
+
+Actual292546 executes target selection, the weapon4/nonzero flag override, damageable688 gate, client/context/count checks and actual33715 mode predicate. Same-team contact in an enabled mode returns zero when friendly-fire34087532 is zero. Other reached client paths stop at292675; nonclient and impact-only branches stop at their actual continuations. The contexts are source-produced original-model hits/misses with exact report hash provenance. Damageable/team/client/weapon flags remain explicit probes, and no damage outcome is invented. Later armor/location/damage and impact side effects remain open.
+
+No original executable/full QVM/native collision/physics/rendering executes. Native responses and legal input domains are supplied or unverified where stated; full runtime/macOS/Linux Rust/Bevy/Solari gameplay remains open.

docs/research/README.md

@@ -699,3 +699,6 @@ The extension collector adds five focused tests for native tail calls, command-r
 
 
 [RE-699: Original-asset numerical geometry comparison audit](MD3_GEOMETRY_AUDIT.md). Reproduce with `python3 scripts/analyze_md3_geometry_audit.py`; check with `python3 scripts/test_md3_geometry_audit.py`.
+
+
+[RE-700: Contact eligibility gates with original-model hit contexts](QVM_CONTACT_GATES.md). Reproduce with `python3 scripts/analyze_qvm_contact_gates.py`; check with `python3 scripts/test_qvm_contact_gates.py`.

scripts/analyze_qvm_contact_gates.py

@@ -0,0 +1,32 @@
+#!/usr/bin/env python3
+"""292545 eligibility prefix with source-produced original-model hit contexts."""
+import hashlib,itertools,json
+from pathlib import Path
+from qvm_research_support import source
+from qvm_trace_support import resume
+from analyze_qvm_victim_stats import mode
+from analyze_qvm_entity_init import POOL,STRIDE
+from analyze_qvm import write_json
+LOCAL=0x79c00000
+ATTACKER=LOCAL+4096
+ATTACKER_CLIENT=LOCAL+8192
+VICTIM_CLIENT=LOCAL+12288
+TRACE=LOCAL+16384
+CONTEXT=LOCAL+20480
+TARGET=POOL+65*STRIDE
+
+def gates(ins,m,args):
+    m=dict(m)
+    for o,v in args.items():m[LOCAL+124+o]=v
+    def predicate(i,m,a):m,r=mode(i,m,a[8]);return m,r['result'],r
+    m,e=resume(ins,292546,293730,m,LOCAL,handlers={33715:predicate},boundaries=[292675,293362,293630]);status='return-zero' if e['return']==0 else {292675:'eligible-client-continuation',293362:'nonclient-continuation',293630:'impact-only-continuation'}[e['stop']]
+    return m,{'target_pointer':m[LOCAL+48],'special_flag':m[LOCAL+52],'status':status,'execution':e}
+
+def main():
+    q,raw,h,ins=source();path=Path('docs/research/md3-original-model-hit.json');data=path.read_bytes();models=json.loads(data)['probes'];hit=next(i for i,p in enumerate(models) if p['context'][16]);miss=next(i for i,p in enumerate(models) if not p['context'][16]);rows=[]
+    for context_kind,damageable,special,client,game,same,friendly in itertools.product(['none','miss','hit'],[0,1],[False,True],[False,True],[2,3,9],[False,True],[0,1]):
+        source_row=hit if context_kind=='hit' else miss;context=models[source_row]['context'];m={TRACE+52:65,TARGET+688:damageable,TARGET+520:VICTIM_CLIENT if client else 0,ATTACKER+520:ATTACKER_CLIENT,ATTACKER_CLIENT+3096:1,VICTIM_CLIENT+3096:1 if same else 2,34090788:game,34087532:friendly}
+        for j,v in enumerate(context):m[CONTEXT+4*j]=v
+        before=[m[TARGET+688],m[TARGET+520],m[ATTACKER+520]];m,r=gates(ins,m,{8:ATTACKER,12:TRACE,16:4 if special else 3,20:0 if context_kind=='none' else CONTEXT,24:int(special),28:0});r.update({'input':{'context_kind':context_kind,'context_source_row':source_row,'context_count':0 if context_kind=='none' else context[16],'damageable':damageable,'special':special,'client':client,'mode':game,'same_team':same,'friendly_fire':friendly},'external_entity_fields_unchanged':before==[m[TARGET+688],m[TARGET+520],m[ATTACKER+520]],'context_unchanged':all(m[CONTEXT+4*j]==v for j,v in enumerate(context))});rows.append(r)
+    write_json(Path('docs/research/qvm-contact-gates.json'),{'source':q['source'],'sha256':q['sha256'],'context_report':str(path),'context_report_sha256':hashlib.sha256(data).hexdigest(),'probes':rows,'limits':'Actual292546 prefix and actual33715 mode predicate using cached source-produced original-model contexts. Client/nonclient/damageable/team flags are explicit branch fixtures. Stops before damage/impact/client continuation; no damage result supplied.'})
+if __name__=='__main__':main()

scripts/test_qvm_contact_gates.py

@@ -0,0 +1,20 @@
+import hashlib,json,unittest
+from pathlib import Path
+class ContactGateTests(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):cls.r=json.loads(Path('docs/research/qvm-contact-gates.json').read_text())
+    def test_independent_gate_truth_table(self):
+        self.assertEqual(len(self.r['probes']),288)
+        for p in self.r['probes']:
+            a=p['input'];expected='eligible-client-continuation'
+            if not a['damageable'] and not a['special']:expected='impact-only-continuation'
+            elif not a['client']:expected='nonclient-continuation'
+            elif not a['context_count'] or (a['mode']==3 and a['same_team'] and not a['friendly_fire']):expected='return-zero'
+            self.assertEqual(p['status'],expected);self.assertEqual(p['special_flag'],int(a['special']));self.assertEqual(p['target_pointer'],34091368+65*880)
+    def test_mode_predicate_only_after_live_context_gates(self):
+        for p in self.r['probes']:
+            a=p['input'];called=bool((a['damageable'] or a['special']) and a['client'] and a['context_count']);self.assertEqual([c['target'] for c in p['execution']['calls']],[33715] if called else [])
+    def test_source_original_model_contexts_have_exact_identity(self):self.assertEqual(hashlib.sha256(Path(self.r['context_report']).read_bytes()).hexdigest(),self.r['context_report_sha256'])
+    def test_no_external_state_writes_or_damage_calls(self):
+        for p in self.r['probes']:self.assertTrue(p['context_unchanged']);self.assertTrue(p['external_entity_fields_unchanged']);self.assertNotIn(211239,[c['target'] for c in p['execution']['calls']])
+if __name__=='__main__':unittest.main()