axon-lang 2.33.0

AXON — the formal cognitive language: a deterministic, proof-carrying AI runtime. Native Rust lexer/parser/type-checker/IR generator (re-exported from axon-frontend) plus the runtime: typed channels (π-calculus mobility, capability extrusion), algebraic effects via Free Monad CPS handlers, lease kernel + reconcile loop, the Epistemic Security Kernel, Trust Types, Proof-Carrying Code (independently verifiable proof objects), and the closed-catalog extension mechanism. Crate publishes as `axon-lang`; library import is `use axon::*` so existing call sites keep working unchanged.
Documentation
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
//! §Fase 51.a — the INDEPENDENT proof checker (the consumer side).
//!
//! This is the trust boundary of Proof-Carrying Code. A consumer who
//! does NOT trust the axon compiler runs [`check_proof`] against the
//! artifact + the proof term and gets an unforgeable verdict. The
//! checker is small, total, and never-panics on arbitrary proof input
//! — it is the only piece a verifier must audit.
//!
//! ## D51.2 — independence (the soundness contract)
//!
//! The checker RE-DERIVES the property from the artifact and verifies
//! the witness against that re-derivation. It NEVER accepts a witness
//! field on faith. A forged witness — one claiming `shield_present:
//! true` for an endpoint whose `shield_ref` is empty, or omitting a
//! phantom class from `unknown_classes` — is rejected because the
//! checker recomputes those facts from the IR and compares. The
//! adversarial tests (`forged_*_rejected`) are first-class gates.
//!
//! ## D51.1 — digest binding
//!
//! Before checking any property the checker recomputes the artifact
//! digest and rejects a mismatch ([`CheckOutcome::DigestMismatch`]).
//! A proof minted for program A cannot be replayed against program B.

use crate::ir_nodes::IRProgram;

use super::generate::{
    artifact_digest, derive_aggregate_soundness_witnesses,
    derive_capability_containment_witness, derive_capability_isolation_witness,
    derive_compliance_coverage_witness, derive_channel_delivery_soundness_witness,
    derive_effect_budgeted_witness, derive_effect_row_soundness_witness,
    derive_endpoint_retry_witness, derive_json_shape_soundness_witness,
    derive_shield_halt_witness, derive_socket_credit_witness, derive_tool_call_soundness_witness,
};
use super::proof_term::{ProofTerm, PropertyClass, ResourceBoundsWitness, Witness};

/// The closed verdict catalog. Total over every proof shape — no
/// default / panic path.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum CheckOutcome {
    /// The property holds for this artifact and the witness matches the
    /// checker's independent re-derivation.
    Verified,
    /// The witness matches the artifact but the property does NOT hold
    /// (an honest defect — e.g. compliance declared with no resolvable
    /// shield, or a phantom regulatory class), OR the witness DISAGREES
    /// with the artifact (a forged proof). `reason` distinguishes.
    Refuted { reason: String },
    /// The proof's `artifact_digest` does not match the supplied
    /// artifact — the proof is about a different program.
    DigestMismatch,
    /// The proof's property/witness pairing is not one this checker
    /// version understands (forward-compat for §51.b-e classes).
    UnknownProperty,
}

/// §51.a — verify a [`ProofTerm`] against the artifact it claims to be
/// about. Independent of the producer (D51.2). Total + never-panics.
pub fn check_proof(proof: &ProofTerm, ir: &IRProgram) -> CheckOutcome {
    // (1) D51.1 — digest binding. Recompute; reject a mismatch before
    // looking at the witness at all.
    if proof.artifact_digest != artifact_digest(ir) {
        return CheckOutcome::DigestMismatch;
    }
    // (2) dispatch on the (property, witness) pairing. A mismatched
    // pairing (e.g. property says ComplianceCoverage but the witness is
    // an EffectRowSoundness witness) is a malformed proof — neither
    // property is actually witnessed — so it is UnknownProperty, not a
    // silent accept.
    match (&proof.property, &proof.witness) {
        (PropertyClass::ComplianceCoverage, Witness::ComplianceCoverage(w)) => {
            check_compliance_coverage(w, ir)
        }
        (PropertyClass::EffectRowSoundness, Witness::EffectRowSoundness(w)) => {
            check_effect_row_soundness(w, ir)
        }
        (PropertyClass::CapabilityIsolation, Witness::CapabilityIsolation(w)) => {
            check_capability_isolation(w, ir)
        }
        (PropertyClass::ResourceBounds, Witness::ResourceBounds(w)) => {
            check_resource_bounds(w, ir)
        }
        (PropertyClass::ShieldHaltGuarantee, Witness::ShieldHaltGuarantee(w)) => {
            check_shield_halt_guarantee(w, ir)
        }
        (PropertyClass::CapabilityContainment, Witness::CapabilityContainment(w)) => {
            check_capability_containment(w, ir)
        }
        (PropertyClass::ToolCallSoundness, Witness::ToolCallSoundness(w)) => {
            check_tool_call_soundness(w, ir)
        }
        (PropertyClass::EffectBudgeted, Witness::EffectBudgeted(w)) => {
            check_effect_budgeted(w, ir)
        }
        (PropertyClass::JsonShapeSoundness, Witness::JsonShapeSoundness(w)) => {
            check_json_shape_soundness(w, ir)
        }
        (PropertyClass::ChannelDeliverySoundness, Witness::ChannelDeliverySoundness(w)) => {
            check_channel_delivery_soundness(w, ir)
        }
        (PropertyClass::AggregateSoundness, Witness::AggregateSoundness(w)) => {
            check_aggregate_soundness(w, ir)
        }
        _ => CheckOutcome::UnknownProperty,
    }
}

/// §52.a — the result of checking one proof inside a bundle.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct ProofCheck {
    /// Position of the proof in `ProofBundle::proofs`.
    pub index: usize,
    /// The property class the proof claims.
    pub property: PropertyClass,
    /// The witness subject (endpoint / tool / store / shield name).
    pub subject: String,
    /// The independent checker's verdict.
    pub outcome: CheckOutcome,
}

/// §52.a — the deployability report for a whole [`ProofBundle`].
///
/// Aggregates the per-proof [`check_proof`] outcomes so a consumer (the
/// enterprise deploy gate, the `axon pcc verify` CLI) has ONE trusted
/// predicate for "is this bundle deployable." The policy — what counts
/// as deployable — lives HERE, on the checker side, never in consumer
/// glue (D52.1): a bundle is deployable iff every proof `Verified`.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct BundleReport {
    /// One entry per proof, in bundle order.
    pub results: Vec<ProofCheck>,
}

impl BundleReport {
    /// True iff EVERY proof verified. Fail-closed: an empty bundle is
    /// vacuously verified (nothing to refute), and any non-`Verified`
    /// outcome (`Refuted` / `DigestMismatch` / `UnknownProperty`) makes
    /// the whole bundle non-deployable (D52.2).
    pub fn all_verified(&self) -> bool {
        self.results
            .iter()
            .all(|r| r.outcome == CheckOutcome::Verified)
    }

    /// Every check that did NOT verify — the reasons a deploy gate
    /// surfaces when it rejects.
    pub fn refutations(&self) -> Vec<&ProofCheck> {
        self.results
            .iter()
            .filter(|r| r.outcome != CheckOutcome::Verified)
            .collect()
    }
}

/// §52.a — check every proof in a bundle against the artifact,
/// independently (each proof re-derived via [`check_proof`]). The
/// bundle's own `artifact_digest` is advisory only — the authoritative
/// binding is the per-proof digest [`check_proof`] re-verifies, so a
/// bundle whose proofs are about a different artifact reports
/// `DigestMismatch` per proof (not a silent pass).
pub fn check_bundle(bundle: &super::proof_term::ProofBundle, ir: &IRProgram) -> BundleReport {
    let results = bundle
        .proofs
        .iter()
        .enumerate()
        .map(|(index, proof)| ProofCheck {
            index,
            property: proof.property.clone(),
            subject: proof.witness.subject_name().to_string(),
            outcome: check_proof(proof, ir),
        })
        .collect();
    BundleReport { results }
}

/// §51.a — re-derive the compliance-coverage witness from the artifact
/// and verify it against the proof's witness, then render the verdict.
///
/// The order matters for soundness: we recompute the witness FIRST
/// (independent of the proof's claims), reject any disagreement as a
/// forgery, and only THEN decide Verified vs Refuted on the
/// re-derived facts.
fn check_compliance_coverage(
    claimed: &super::proof_term::ComplianceCoverageWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    // Locate the endpoint named in the witness. Absent ⟹ the proof is
    // about an endpoint not in this artifact (forged / stale).
    let ep = match ir.endpoints.iter().find(|e| e.name == claimed.endpoint_name) {
        Some(e) => e,
        None => {
            return CheckOutcome::Refuted {
                reason: format!(
                    "endpoint '{}' not present in artifact",
                    claimed.endpoint_name
                ),
            }
        }
    };

    // Re-derive independently — do NOT trust the witness (D51.2).
    let actual = derive_compliance_coverage_witness(
        &ep.name,
        &ep.compliance,
        &ep.shield_ref,
        ir,
    );

    // Forgery detection: every witness field must equal the checker's
    // re-derivation. Any disagreement ⟹ the proof lies about the
    // artifact.
    if actual != *claimed {
        return CheckOutcome::Refuted {
            reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                .to_string(),
        };
    }

    // Verdict on the re-derived (trusted) facts.
    if !actual.shield_present {
        return CheckOutcome::Refuted {
            reason: format!(
                "endpoint '{}' declares compliance {:?} but has no resolvable shield (shield_ref={:?})",
                actual.endpoint_name, actual.required_classes, actual.shield_ref
            ),
        };
    }
    if !actual.unknown_classes.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "endpoint '{}' declares unknown regulatory class(es) {:?} (not in the closed registry)",
                actual.endpoint_name, actual.unknown_classes
            ),
        };
    }
    if !actual.uncovered_classes.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "endpoint '{}' requires regulatory class(es) {:?} that its shield '{}' does not provide (shield provides {:?})",
                actual.endpoint_name,
                actual.uncovered_classes,
                actual.shield_ref,
                actual.provided_classes
            ),
        };
    }
    CheckOutcome::Verified
}

/// §51.b — re-derive the effect-row witness from the named tool and
/// verify it against the proof, then render the verdict. Independent
/// of the producer (D51.2): the checker re-reads `tool.effect_row` from
/// the artifact and recomputes every fact; a forged witness is rejected
/// because the recomputation disagrees.
fn check_effect_row_soundness(
    claimed: &super::proof_term::EffectRowSoundnessWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    let tool = match ir.tools.iter().find(|t| t.name == claimed.tool_name) {
        Some(t) => t,
        None => {
            return CheckOutcome::Refuted {
                reason: format!(
                    "tool '{}' not present in artifact",
                    claimed.tool_name
                ),
            }
        }
    };

    // Re-derive independently — do NOT trust the witness. §Fase 53.d:
    // the extension provenance members are re-derived from the SAME
    // artifact (`ir`), so an extension-declared base verifies, while a
    // base no extension declares still refutes. Soundness invariant #1:
    // the verifier reads the artifact's own `extensions`, never an
    // external registry.
    let ext_members = super::generate::extension_effect_members(ir);
    let actual =
        derive_effect_row_soundness_witness(&tool.name, &tool.effect_row, &ext_members);

    if actual != *claimed {
        return CheckOutcome::Refuted {
            reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                .to_string(),
        };
    }

    // Verdict on the re-derived facts.
    if !actual.unknown_bases.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "tool '{}' declares effect entr(ies) with unknown base(s) {:?} (not in the closed effect catalog)",
                actual.tool_name, actual.unknown_bases
            ),
        };
    }
    if !actual.missing_qualifier.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "tool '{}' declares qualifier-requiring effect(s) {:?} without a qualifier (bare stream/trust is unenforceable)",
                actual.tool_name, actual.missing_qualifier
            ),
        };
    }
    if !actual.invalid_stream_qualifier.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "tool '{}' declares stream effect(s) {:?} with an invalid backpressure policy qualifier",
                actual.tool_name, actual.invalid_stream_qualifier
            ),
        };
    }
    if actual.purity_violation {
        return CheckOutcome::Refuted {
            reason: format!(
                "tool '{}' declares `pure` alongside other effects {:?} — a pure tool cannot be effectful",
                actual.tool_name, actual.declared_effects
            ),
        };
    }
    CheckOutcome::Verified
}

/// §51.c — re-derive the capability-gate witness from the named store
/// and verify it against the proof, then render the verdict.
/// Independent of the producer (D51.2): the checker re-reads
/// `store.capability` from the artifact and re-runs the §32.g grammar
/// validator; a forged witness (claiming `malformed: false` for a
/// broken gate) is rejected because the recomputation disagrees.
fn check_capability_isolation(
    claimed: &super::proof_term::CapabilityIsolationWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    let store = match ir
        .axonstore_specs
        .iter()
        .find(|s| s.name == claimed.store_name)
    {
        Some(s) => s,
        None => {
            return CheckOutcome::Refuted {
                reason: format!(
                    "axonstore '{}' not present in artifact",
                    claimed.store_name
                ),
            }
        }
    };

    // Re-derive independently — do NOT trust the witness.
    let actual = derive_capability_isolation_witness(&store.name, &store.capability);

    if actual != *claimed {
        return CheckOutcome::Refuted {
            reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                .to_string(),
        };
    }

    if actual.malformed {
        return CheckOutcome::Refuted {
            reason: format!(
                "axonstore '{}' declares a malformed capability gate {:?} (not a valid §32.g scope: ^[a-z][a-z0-9_]*(\\.[a-z][a-z0-9_]*)*$)",
                actual.store_name, actual.capability
            ),
        };
    }
    CheckOutcome::Verified
}

/// §51.d — re-derive the resource-bound witness from the named subject
/// (endpoint or socket) and verify it against the proof, then render
/// the verdict. Independent of the producer (D51.2): the checker
/// re-reads `retries` / `backpressure_credit` from the artifact and
/// recomputes the bound; a forged witness is rejected because the
/// recomputation disagrees.
fn check_resource_bounds(
    claimed: &ResourceBoundsWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    match claimed {
        ResourceBoundsWitness::EndpointRetry { endpoint_name, .. } => {
            let ep = match ir.endpoints.iter().find(|e| e.name == *endpoint_name) {
                Some(e) => e,
                None => {
                    return CheckOutcome::Refuted {
                        reason: format!(
                            "endpoint '{}' not present in artifact",
                            endpoint_name
                        ),
                    }
                }
            };
            let actual = derive_endpoint_retry_witness(&ep.name, ep.retries);
            if actual != *claimed {
                return CheckOutcome::Refuted {
                    reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                        .to_string(),
                };
            }
            if let ResourceBoundsWitness::EndpointRetry { retries, in_bounds, .. } = &actual {
                if !in_bounds {
                    return CheckOutcome::Refuted {
                        reason: format!(
                            "endpoint '{}' declares retries={} outside the bound [0, {}] (negative is nonsensical; above the ceiling is a retry storm)",
                            ep.name,
                            retries,
                            super::proof_term::MAX_RETRIES
                        ),
                    };
                }
            }
            CheckOutcome::Verified
        }
        ResourceBoundsWitness::SocketCredit { socket_name, .. } => {
            let socket = match ir.sockets.iter().find(|s| s.name == *socket_name) {
                Some(s) => s,
                None => {
                    return CheckOutcome::Refuted {
                        reason: format!(
                            "socket '{}' not present in artifact",
                            socket_name
                        ),
                    }
                }
            };
            // The witness claims a DECLARED credit. If the artifact's
            // socket has no declared credit, the witness disagrees
            // (forged / stale).
            let credit = match socket.backpressure_credit {
                Some(k) => k,
                None => {
                    return CheckOutcome::Refuted {
                        reason: format!(
                            "socket '{}' has no declared backpressure credit in the artifact, but the witness claims one (forged or stale proof)",
                            socket_name
                        ),
                    }
                }
            };
            let actual = derive_socket_credit_witness(&socket.name, credit);
            if actual != *claimed {
                return CheckOutcome::Refuted {
                    reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                        .to_string(),
                };
            }
            if let ResourceBoundsWitness::SocketCredit { credit, positive, .. } = &actual {
                if !positive {
                    return CheckOutcome::Refuted {
                        reason: format!(
                            "socket '{}' declares backpressure credit({}) — a window < 1 deadlocks the §Fase 41.b typed-resource gate",
                            socket.name, credit
                        ),
                    };
                }
            }
            CheckOutcome::Verified
        }
    }
}

/// §51.e — re-derive the shield breach-policy witness from the named
/// shield and verify it against the proof, then render the verdict.
/// Independent of the producer (D51.2): the checker re-reads
/// `on_breach` + `scan` from the artifact and recomputes both facts; a
/// forged witness (claiming `vacuous_halt: false` for a halt shield
/// that scans nothing) is rejected because the recomputation disagrees.
fn check_shield_halt_guarantee(
    claimed: &super::proof_term::ShieldHaltGuaranteeWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    let shield = match ir.shields.iter().find(|s| s.name == claimed.shield_name) {
        Some(s) => s,
        None => {
            return CheckOutcome::Refuted {
                reason: format!(
                    "shield '{}' not present in artifact",
                    claimed.shield_name
                ),
            }
        }
    };

    // Re-derive independently — do NOT trust the witness.
    let actual =
        derive_shield_halt_witness(&shield.name, &shield.on_breach, &shield.scan);

    if actual != *claimed {
        return CheckOutcome::Refuted {
            reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                .to_string(),
        };
    }

    if !actual.known_policy {
        return CheckOutcome::Refuted {
            reason: format!(
                "shield '{}' declares an unknown on_breach policy {:?} (not in the closed breach-policy catalog)",
                actual.shield_name, actual.on_breach
            ),
        };
    }
    if actual.vacuous_halt {
        return CheckOutcome::Refuted {
            reason: format!(
                "shield '{}' declares `on_breach: halt` but an empty `scan: []` — the halt can never fire (no scan ⟹ no breach ⟹ no halt): a vacuous guarantee",
                actual.shield_name
            ),
        };
    }
    CheckOutcome::Verified
}

/// §51.x — re-derive the capability-containment witness from the named
/// endpoint and verify it against the proof, then render the verdict.
/// Independent of the producer (D51.2): the checker re-resolves the
/// `execute_flow`, re-walks its reachable store ops, re-resolves each
/// gate, and recomputes the uncovered set; a forged witness (e.g.
/// hiding an uncovered gate) is rejected because the recomputation
/// disagrees.
fn check_capability_containment(
    claimed: &super::proof_term::CapabilityContainmentWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    let ep = match ir.endpoints.iter().find(|e| e.name == claimed.endpoint_name) {
        Some(e) => e,
        None => {
            return CheckOutcome::Refuted {
                reason: format!(
                    "endpoint '{}' not present in artifact",
                    claimed.endpoint_name
                ),
            }
        }
    };

    // Re-derive independently — do NOT trust the witness.
    let actual = derive_capability_containment_witness(
        &ep.name,
        &ep.execute_flow,
        &ep.requires_capabilities,
        ir,
    );

    if actual != *claimed {
        return CheckOutcome::Refuted {
            reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                .to_string(),
        };
    }

    // Verdict. An unresolvable execute_flow cannot be analyzed for
    // store reachability — refuse to certify containment (sound: a
    // missing flow could reach gated stores we cannot see).
    if !actual.flow_resolved {
        return CheckOutcome::Refuted {
            reason: format!(
                "endpoint '{}' executes flow '{}' which is not present in the artifact — cannot certify capability containment",
                actual.endpoint_name, actual.execute_flow
            ),
        };
    }
    if !actual.uncovered_gates.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "endpoint '{}' reaches store(s) gated by capabilit(ies) {:?} that its declared `requires:` {:?} does not cover — a capability leak (a request satisfying the declared requires could reach a store it is not authorized for)",
                actual.endpoint_name, actual.uncovered_gates, actual.declared_requires
            ),
        };
    }
    CheckOutcome::Verified
}

/// §58.i — re-derive the tool-call-soundness witness from the named flow
/// + call site and verify it against the proof, then render the verdict.
/// Independent of the producer (D51.2): the checker re-walks the SAME
/// digest-bound IR, locates the call at `call_index`, re-reads the called
/// tool's `parameters:` schema, and recomputes every fact; a forged
/// witness (e.g. hiding an unknown argument or a type mismatch) is
/// rejected because the recomputation disagrees.
fn check_tool_call_soundness(
    claimed: &super::proof_term::ToolCallSoundnessWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    // Re-derive independently — do NOT trust the witness. `None` ⟹ the
    // flow or the call site named by the witness is absent from this
    // artifact (forged / stale proof).
    let actual = match derive_tool_call_soundness_witness(
        &claimed.flow_name,
        claimed.call_index,
        ir,
    ) {
        Some(w) => w,
        None => {
            return CheckOutcome::Refuted {
                reason: format!(
                    "flow '{}' has no structured `use` call at index {} in this artifact",
                    claimed.flow_name, claimed.call_index
                ),
            }
        }
    };

    if actual != *claimed {
        return CheckOutcome::Refuted {
            reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                .to_string(),
        };
    }

    // Verdict on the re-derived (trusted) facts.
    if !actual.schema_present {
        // A schema-less / undeclared tool carries no arg contract — there
        // is nothing to certify, so a proof for it is vacuously verified
        // (generation never emits one; this stays total for a hand-built
        // witness).
        return CheckOutcome::Verified;
    }
    if !actual.unknown_args.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "call to tool '{}' in flow '{}' passes argument(s) {:?} the tool does not declare (declared parameters: {:?})",
                actual.tool_name, actual.flow_name, actual.unknown_args, actual.declared_params
            ),
        };
    }
    if !actual.duplicate_args.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "call to tool '{}' in flow '{}' supplies duplicate argument(s) {:?}",
                actual.tool_name, actual.flow_name, actual.duplicate_args
            ),
        };
    }
    if !actual.missing_required.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "call to tool '{}' in flow '{}' omits required argument(s) {:?}",
                actual.tool_name, actual.flow_name, actual.missing_required
            ),
        };
    }
    if !actual.type_mismatches.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "call to tool '{}' in flow '{}' has literal-argument type mismatch(es) {:?} (each `arg:expected:got`)",
                actual.tool_name, actual.flow_name, actual.type_mismatches
            ),
        };
    }
    CheckOutcome::Verified
}

/// §72.f — check an [`EffectBudgetedWitness`]. Re-derive the budget facts
/// independently from the artifact; refute on disagreement (forgery) or on any
/// budget defect (an unresolved effect, a non-positive limit, an invalid period,
/// or an out-of-catalog `on_exhausted`).
fn check_effect_budgeted(
    claimed: &super::proof_term::EffectBudgetedWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    // Re-derive — do NOT trust the witness. `None` ⟹ the daemon named by the
    // witness is absent, or it carries no budget (forged / stale proof).
    let actual = match derive_effect_budgeted_witness(&claimed.daemon_name, ir) {
        Some(w) => w,
        None => {
            return CheckOutcome::Refuted {
                reason: format!(
                    "daemon '{}' has no `budget {{ }}` in this artifact",
                    claimed.daemon_name
                ),
            }
        }
    };
    if actual != *claimed {
        return CheckOutcome::Refuted {
            reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                .to_string(),
        };
    }
    // Verdict on the re-derived (trusted) facts.
    if !actual.unresolved_effects.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "daemon '{}' budget targets undefined tool(s) {:?} — no matching `tool` declaration",
                actual.daemon_name, actual.unresolved_effects
            ),
        };
    }
    if !actual.nonpositive_limits.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "daemon '{}' budget has non-positive quota limit(s) {:?} (each `effect:kind`)",
                actual.daemon_name, actual.nonpositive_limits
            ),
        };
    }
    if !actual.invalid_periods.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "daemon '{}' budget has out-of-catalog period(s) {:?} (each `effect:period`; valid: {:?})",
                actual.daemon_name, actual.invalid_periods, super::proof_term::VALID_BUDGET_PERIODS
            ),
        };
    }
    if !actual.on_exhausted_valid {
        return CheckOutcome::Refuted {
            reason: format!(
                "daemon '{}' budget has an out-of-catalog `on_exhausted` policy '{}' (valid: {:?})",
                actual.daemon_name, actual.on_exhausted, super::proof_term::VALID_ON_EXHAUSTED
            ),
        };
    }
    CheckOutcome::Verified
}

/// §73.g — check a [`JsonShapeSoundnessWitness`]. Re-derive the store's
/// lens shapes independently from the artifact; refute on disagreement
/// (forgery) or on any shape that does not resolve to a declared struct
/// `type` (the `axon-T840` invariant, now independently verified).
fn check_json_shape_soundness(
    claimed: &super::proof_term::JsonShapeSoundnessWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    // Re-derive — do NOT trust the witness. `None` ⟹ the store named by the
    // witness is absent, or carries no inline `Json<T>` lens column (a
    // forged / stale proof).
    let actual = match derive_json_shape_soundness_witness(&claimed.store_name, ir) {
        Some(w) => w,
        None => {
            return CheckOutcome::Refuted {
                reason: format!(
                    "axonstore '{}' declares no inline `Json<T>` lens column in this artifact",
                    claimed.store_name
                ),
            }
        }
    };
    if actual != *claimed {
        return CheckOutcome::Refuted {
            reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                .to_string(),
        };
    }
    // Verdict on the re-derived (trusted) facts: every lens shape must
    // resolve to a declared struct `type`.
    if !actual.unresolved_shapes.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "axonstore '{}' declares `Json<T>` lens column(s) {:?} (each `column:Shape`) \
                 whose shape is not a declared `type` — the lens cannot be field-checked, so \
                 navigation soundness is unprovable. Declare the struct, or use open `Json`.",
                actual.store_name, actual.unresolved_shapes
            ),
        };
    }
    CheckOutcome::Verified
}

/// §74.g — check a [`ChannelDeliverySoundnessWitness`]. Re-derive the
/// channel's producer/consumer facts from the artifact; refute on
/// disagreement (forgery) or when a consumed channel has NO producer (a
/// `listen`er that can never fire — nothing `emit`s to it).
fn check_channel_delivery_soundness(
    claimed: &super::proof_term::ChannelDeliverySoundnessWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    let actual = match derive_channel_delivery_soundness_witness(&claimed.channel_name, ir) {
        Some(w) => w,
        None => {
            return CheckOutcome::Refuted {
                reason: format!(
                    "channel '{}' has no `daemon` `listen`er in this artifact (no delivery \
                     contract to certify)",
                    claimed.channel_name
                ),
            }
        }
    };
    if actual != *claimed {
        return CheckOutcome::Refuted {
            reason: "witness disagrees with artifact re-derivation (forged or stale proof)"
                .to_string(),
        };
    }
    // Verdict: a consumed channel with no producer never fires.
    if !actual.has_producer {
        return CheckOutcome::Refuted {
            reason: format!(
                "channel '{}' has a `daemon` `listen`er but NO producer — nothing `emit`s to \
                 it, so the listener can never fire. Add an `emit {}(…)` in a flow, or remove \
                 the listener (the Kivi brief #39 defect, now machine-checked).",
                actual.channel_name, actual.channel_name
            ),
        };
    }
    CheckOutcome::Verified
}

/// §76.e — check an [`AggregateSoundnessWitness`]. Re-derive ALL aggregate
/// sites from the artifact (through the SAME runtime clause parser the
/// engines execute) and require the claimed witness to be one of them —
/// a witness about a site the artifact does not contain is a forgery.
/// Then the verdict: any recorded violation refutes (the rendered SQL
/// would not aggregate exactly the declared columns through the closed
/// catalog).
fn check_aggregate_soundness(
    claimed: &super::proof_term::AggregateSoundnessWitness,
    ir: &IRProgram,
) -> CheckOutcome {
    let derived = derive_aggregate_soundness_witnesses(ir);
    if !derived.iter().any(|w| w == claimed) {
        return CheckOutcome::Refuted {
            reason: format!(
                "no aggregate-retrieve site in this artifact matches the witness \
                 (flow '{}', store '{}', aggregate '{}') — forged or stale proof",
                claimed.flow_name, claimed.store_name, claimed.aggregate
            ),
        };
    }
    if !claimed.violations.is_empty() {
        return CheckOutcome::Refuted {
            reason: format!(
                "aggregate-retrieve on store '{}' in flow '{}' is UNSOUND: {}",
                claimed.store_name,
                claimed.flow_name,
                claimed.violations.join("; ")
            ),
        };
    }
    CheckOutcome::Verified
}