Clarify caught-primary preservation and accounting requirements

research/h25-g12-original-contract-scope/CLAIMS.json

@@ -0,0 +1,34 @@
+{
+  "schema": "h25-g12-original-contract-scope-v1",
+  "resolved": [
+    {
+      "id": "G12-CAUGHT-PRIMARY",
+      "input": "research/h25-cache-source-handoff/snapshot/review/cache-s0308-source-completion-v2/g12-protocol-independent/G12-CANDIDATE-PROTOCOL-REVIEW-v1.json",
+      "pointer": "/blocking_findings/0/required_narrow_change",
+      "source_text": "Protect all newly added secondary bookkeeping that runs while preserving an already-caught P, particularly the last handler at476-477, so failures there cannot escape in place of P. Retain designated secondary errors in pre-existing raw slots before optional allocating journals where required; keep optional reporting guarded. No broader allocation-equivalence proof or new authority framework is required.",
+      "interpretation": "This bounded repair protects an already-caught primary against newly added bookkeeping. No new universal allocation-equivalence obligation is inferred."
+    },
+    {
+      "id": "G12-SOURCE-BOUNDARY",
+      "input": "research/h25-cache-source-handoff/snapshot/review/cache-s0308-source-completion-v2/g12-protocol-independent/G12-CANDIDATE-PROTOCOL-REVIEW-v2.json",
+      "pointer": "/finding_disposition/0/boundary",
+      "source_text": "This is static preservation of an already-caught primary against the newly added bookkeeping path. It is not a universal allocation-equivalence proof, actual injected-failure run, or complete original cleanup qualification.",
+      "interpretation": "The accepted narrow static repair is distinct from injected runtime failure and complete cleanup qualification."
+    },
+    {
+      "id": "G12-NO-NEW-FINANCIAL-API",
+      "input": "research/h25-cache-source-handoff/snapshot/review/cache-s0308-source-completion-v2/g12-custody-independent/G12-CUSTODY-SOURCE-SCOPE-CLARIFICATION-v1.json",
+      "pointer": "/answer",
+      "source_text": "The inspected original SC5/SC6 checklist and cost/caller contracts do not require construction of a new enclosing financial/accounting API for reference custody. They require authentic independent original reference transport, source and alias ownership, and a complete source work/cost bill preserving actual measurement, limits, all tails and unmeasured quantities. No exact original causal dependency on new F4/M source work was demonstrated. The additional final-publication/enclosing-completion API is a requirement introduced by the G12 design as written; its absence is not by itself a justified original S03.08 source-completion blocker.",
+      "interpretation": "A newly constructed enclosing accounting API is not an original prerequisite. Authentic transport, ownership, complete work/cost bills, measurements and unknown tails remain required."
+    }
+  ],
+  "conditional_PR6347_witnesses_withdrawn": false,
+  "new_runtime_or_activity_acceptance": false,
+  "original_weighted_activity_percent": 43.5,
+  "original_complete_activities": 22,
+  "original_activity_denominator": 49,
+  "stage_gate_percent": 20,
+  "overall_milestone_percent": 10,
+  "costs": "OPEN_NOT_ZERO"
+}

research/h25-g12-original-contract-scope/README.md

@@ -0,0 +1,30 @@
+# Original G12 contract scope
+
+The retained original protocol reviews require preservation of an **already-caught primary** against newly added bookkeeping failures. They explicitly do not demand a broader allocation-equivalence proof or new authority framework for that bounded source repair. The retained custody clarification likewise does not make a newly constructed enclosing financial/accounting API an original S03.08 prerequisite.
+
+This resolves two scope questions. It does not repair or accept the remaining integration. Authentic independent reference transport, first-primary/raw-secondary identities, closed aliases, cleanup, source/mode/output custody, terminal order, a complete work/cost bill and unknown tails remain original obligations. Neither the original checklist nor a design-specific predicate is silently deleted. Any change to a design that added an enclosing completion API needs its own explicit successor and review.
+
+## Evidence and relation to existing work
+
+`CLAIMS.json` preserves three exact source texts and JSON pointers. `SOURCES.json` binds the unchanged live canonical review files at base `7c49fe32c55de02c2c873655c804b16e4c3fef0a`; `VERIFY.json` declares all three for invalidation. The verifier pins bytes and checks pointer transcription, with positive and tamper controls. The written interpretation remains an inspectable source argument; the verifier is not a natural-language proof engine or an exception-safety test.
+
+[PR6321](https://github.com/CrocSwap/integer-mult-bounds/pull/6321) retains these original reviews. [PR6339](https://github.com/CrocSwap/integer-mult-bounds/pull/6339) addresses identity/ordinary cleanup; [PR6346](https://github.com/CrocSwap/integer-mult-bounds/pull/6346) adds narrow custody repairs. [PR6347](https://github.com/CrocSwap/integer-mult-bounds/pull/6347) gives conditional acquisition/raw-error/delivery witnesses. Those witnesses are not withdrawn; they must not be promoted into an additional universal prerequisite for the narrower original source review. This contribution neither requires nor replays the unmerged PR6346/6347 packages.
+
+## Reproduce the bounded audit
+
+```sh
+python3 -B scripts/research_package.py --package research/h25-g12-original-contract-scope --check-only
+python3 -B scripts/research_package.py --package research/h25-g12-original-contract-scope
+```
+
+Both registered normal/optimized commands inspect JSON and source bytes only. No archived target, Root, provider, compiler, native helper or original control is imported or executed. A passing replay is evidence-binding, not runtime qualification or independent human/proof-kernel acceptance.
+
+Original weighted completion remains43.5% (22/49), stage gates20%, milestones10%; no new body/family/activity, source adoption, O03, gain or kappa credit. All original selections remainNone. Unknown physical, preparation, review, publication and final metadata costs remain `OPEN_NOT_ZERO`.
+
+## Open next questions
+
+Review a concrete successor against the original obligations; see `TASK.json`. The smallest step is an exact caught-primary/secondary/cleanup source review, without adding a universal interpreter premise or generic accounting framework.
+
+## Credit
+
+Chris R / Clear1nsight, with substantial OpenAI Codex and ChatGPT Pro assistance. Retain the original PR6321/6029/5864 authors and source reviewers, unpaidmoon with Cursor Grok assistance for PR6339, and Douglas R. Colkitt's circuit/network/verification lineage. Canonical predecessor files remain unchanged. New analysis and verifier: Apache-2.0.

research/h25-g12-original-contract-scope/SOURCES.json

@@ -0,0 +1,36 @@
+{
+  "schema_version": 1,
+  "base_commit": "7c49fe32c55de02c2c873655c804b16e4c3fef0a",
+  "files": [
+    {
+      "path": "certificates/selected-result.json",
+      "usage": "provenance",
+      "git_blob": "d59b5035e7b4cbaa9d688a1d32dd0d6eb79c466a",
+      "sha256": "9b8d0eea95b3bf0dfcb7938a732a6439b87a2685dec2a5ab9e525b3b34debbfa"
+    },
+    {
+      "path": "docs/research/agent-task-catalog.json",
+      "usage": "provenance",
+      "git_blob": "3404fbdd5e7529a0674eef26114b5071e4586cec",
+      "sha256": "ead6cdc9d0af6d58e8a49744d8ec8c7c15e13d2ba93c276c434bcf9b51b93a81"
+    },
+    {
+      "path": "research/h25-cache-source-handoff/snapshot/review/cache-s0308-source-completion-v2/g12-protocol-independent/G12-CANDIDATE-PROTOCOL-REVIEW-v1.json",
+      "usage": "input",
+      "git_blob": "583dfc3936d478ffba6726470b454d0d99247b37",
+      "sha256": "feecd63be55243fe2a7d07da8ac1f7cfcec6558362e694b82219a1be232ab7bc"
+    },
+    {
+      "path": "research/h25-cache-source-handoff/snapshot/review/cache-s0308-source-completion-v2/g12-protocol-independent/G12-CANDIDATE-PROTOCOL-REVIEW-v2.json",
+      "usage": "input",
+      "git_blob": "80d9b3e7d4d9657cd73875d06727a98d6360d612",
+      "sha256": "9b9043fa83f1a052a6708105431db20efd37b88bc0f23adc36491edd27b867ee"
+    },
+    {
+      "path": "research/h25-cache-source-handoff/snapshot/review/cache-s0308-source-completion-v2/g12-custody-independent/G12-CUSTODY-SOURCE-SCOPE-CLARIFICATION-v1.json",
+      "usage": "input",
+      "git_blob": "66737f17c039d7157233648c46e0192e50e0e2dc",
+      "sha256": "a9261f686d4fbe6ca40258335231a730dc7e3319be7cedf7c39e189a57c7f367"
+    }
+  ]
+}

research/h25-g12-original-contract-scope/TASK.json

@@ -0,0 +1,60 @@
+{
+  "schema_version": 1,
+  "task_id": "portable-replay",
+  "priority_revision": 8,
+  "base_commit": "7c49fe32c55de02c2c873655c804b16e4c3fef0a",
+  "original_card": {
+    "id": "portable-replay",
+    "question": "Can one reproducible environment or artifact-comparison failure be repaired?",
+    "starting_points": [
+      "research/gzip-container-reproducibility/README.md",
+      "docs/ci-verification.md"
+    ],
+    "methods": [
+      "Reproduce one portability failure and add a minimal cross-environment fix.",
+      "Audit one canonical-payload comparison for omitted proof-relevant fields."
+    ],
+    "acceptance_gate": "Preserve immutable source hashes and proof-relevant content; a different compressed byte stream alone is not a mathematical mismatch.",
+    "first_step": {
+      "command": [
+        "python3",
+        "-B",
+        "scripts/research_selected_baseline.py"
+      ],
+      "expected": "PASS_RESEARCH_BASELINE_BINDINGS_ONLY",
+      "inspect": "Resolve the current selected review, separate complex/ordinary BIT/admitted BIT/outer budgets, and compare the card against live handoffs before starting. Historical ceilings require exact source/model equivalence.",
+      "scope": "Source-binding orientation only; not a scientific replay or all-size proof. Materialize selected inputs if this checkout is sparse."
+    },
+    "first_experiment": "Reproduce one specific mismatch or audit one canonical comparison for omitted proof-relevant fields.",
+    "bottleneck_link": "Portable exact checks prevent environment failures from blocking valid contributions.",
+    "allowed_variations": [
+      "One interpreter/version mismatch or one payload field comparison."
+    ],
+    "scope_change_triggers": [
+      "Removing immutable-source pins because generated compressed bytes differ."
+    ],
+    "overlap_terms": [
+      "gzip container",
+      "reproducibility",
+      "canonical payload"
+    ]
+  },
+  "scope_status": "changed",
+  "actual_scope": "Resolve the original G12 review scope: preserve an already-caught primary against new bookkeeping failures; neither a universal interpreter allocation-equivalence proof nor a newly constructed enclosing financial API is imposed by those source reviews. Preserve real original obligations and conditional PR6347 findings.",
+  "scope_change_reason": "Explicit user assignment to publish all resolved P07/G12 and V03 source questions, then consult Pro on C05. The retained card is provenance, not a claim to repair the selected supplier.",
+  "bottleneck_link": "Source-bound verifier qualification prerequisites; no change to the multiplication supplier, selected result or framework assumptions.",
+  "budget_minutes": 60,
+  "stop_rule": "Stop on source mismatch or unresolved interpretation; no target/native/compiler/original-control execution, historical allowance transfer or silent scope change.",
+  "outcome": "partial",
+  "evidence": "README.md, CLAIMS.json, exact source inputs and passive verifier with discriminating tamper controls.",
+  "author": "Chris R / Clear1nsight",
+  "ai_assistance": "ChatGPT Pro source consultation and substantial OpenAI Codex source-audit/verification assistance. These are agent-assisted findings, not human or proof-kernel acceptance.",
+  "open_questions": [
+    {
+      "question": "Does a concrete successor preserve every original caught-primary, raw-secondary, ownership, cleanup and caller requirement?",
+      "why_it_matters": "Removing an invented prerequisite does not accept the rejected integration or complete S03.08.",
+      "first_step": "Review a separate exact successor against original CP01-CP15 and all18 branch obligations; keep source, runtime, cost and adoption dispositions separate.",
+      "blocked_by": "The remaining integrated implementation and original qualification have not been established."
+    }
+  ]
+}

research/h25-g12-original-contract-scope/VALIDATION.json

@@ -0,0 +1,16 @@
+{
+  "schema": "h25-resolved-source-validation-v1",
+  "observed_utc": "2026-11-02T03:05:04.863955+00:00",
+  "python": "3.12.3",
+  "base_commit": "7c49fe32c55de02c2c873655c804b16e4c3fef0a",
+  "metadata_and_pins": "PASS",
+  "registered_normal": "PASS",
+  "registered_optimized": "PASS",
+  "summary": "3 pinned canonical review inputs;6 tamper controls in normal/optimized mode",
+  "source_findings_review": "Complete ChatGPT Pro source responses assessed against exact source by Codex. New verifier is a passive evidence check, not independent human or proof-kernel acceptance.",
+  "target_imports_or_execution": false,
+  "native_or_compiler_work": false,
+  "original_controls_executed": false,
+  "full_family_or_activity_credit": false,
+  "costs": "OPEN_NOT_ZERO"
+}

research/h25-g12-original-contract-scope/VERIFY.json

@@ -0,0 +1,38 @@
+{
+  "schema_version": 1,
+  "status": "ready",
+  "python": "3.12",
+  "requirements": [],
+  "inputs": [
+    "research/h25-cache-source-handoff/snapshot/review/cache-s0308-source-completion-v2/g12-protocol-independent/G12-CANDIDATE-PROTOCOL-REVIEW-v1.json",
+    "research/h25-cache-source-handoff/snapshot/review/cache-s0308-source-completion-v2/g12-protocol-independent/G12-CANDIDATE-PROTOCOL-REVIEW-v2.json",
+    "research/h25-cache-source-handoff/snapshot/review/cache-s0308-source-completion-v2/g12-custody-independent/G12-CUSTODY-SOURCE-SCOPE-CLARIFICATION-v1.json"
+  ],
+  "commands": [
+    {
+      "argv": [
+        "{python}",
+        "-I",
+        "-W",
+        "error",
+        "-B",
+        "{package}/verify.py"
+      ],
+      "timeout_seconds": 30,
+      "checks": "Passive source/byte evidence and tamper controls only; archived modules are never imported, compiled or executed. No runtime/family/activity qualification."
+    },
+    {
+      "argv": [
+        "{python}",
+        "-I",
+        "-O",
+        "-W",
+        "error",
+        "-B",
+        "{package}/verify.py"
+      ],
+      "timeout_seconds": 30,
+      "checks": "Passive source/byte evidence and tamper controls only; archived modules are never imported, compiled or executed. No runtime/family/activity qualification."
+    }
+  ]
+}

research/h25-g12-original-contract-scope/verify.py

@@ -0,0 +1,104 @@
+"""Passive review-text binding. Never execute a predecessor or target module."""
+from pathlib import Path
+import copy
+import hashlib
+import json
+
+HERE = Path(__file__).resolve().parent
+ROOT = HERE.parent.parent
+
+
+def need(condition, message):
+    if condition is not True:
+        raise RuntimeError(message)
+
+
+def pointer(value, path):
+    for key in path.split('/')[1:]:
+        value = value[int(key)] if type(value) is list else value[key]
+    return value
+
+
+def inspect_scope(documents, claims):
+    rows = claims['resolved']
+    need([row['id'] for row in rows] == [
+        'G12-CAUGHT-PRIMARY', 'G12-SOURCE-BOUNDARY', 'G12-NO-NEW-FINANCIAL-API'],
+        'claim identity/order changed')
+    for row in rows:
+        need(pointer(documents[row['input']], row['pointer']) == row['source_text'],
+             'source text transcription differs')
+    need('already-caught P' in rows[0]['source_text'] and
+         'No broader allocation-equivalence proof or new authority framework is required.' in rows[0]['source_text'],
+         'original narrow review instruction missing')
+    need('already-caught primary' in rows[1]['source_text'] and
+         'not a universal allocation-equivalence proof' in rows[1]['source_text'],
+         'static/runtime source boundary missing')
+    need('do not require construction of a new enclosing financial/accounting API' in rows[2]['source_text'] and
+         'complete source work/cost bill' in rows[2]['source_text'],
+         'original financial/API distinction missing')
+    need(claims['conditional_PR6347_witnesses_withdrawn'] is False and
+         claims['new_runtime_or_activity_acceptance'] is False,
+         'unearned acceptance or witness retraction')
+    need(claims['original_weighted_activity_percent'] == 43.5 and
+         claims['original_complete_activities'] == 22 and
+         claims['original_activity_denominator'] == 49 and
+         claims['stage_gate_percent'] == 20 and claims['overall_milestone_percent'] == 10,
+         'original denominators/credit changed')
+    need(claims['costs'] == 'OPEN_NOT_ZERO', 'unknown costs erased')
+
+
+def rejected(action, label):
+    try:
+        action()
+    except (RuntimeError, KeyError, IndexError):
+        return
+    raise RuntimeError('tamper accepted: ' + label)
+
+
+def main():
+    sources = json.loads((HERE / 'SOURCES.json').read_text())
+    docs = {}
+    for row in sources['files']:
+        if row['usage'] != 'input':
+            continue
+        raw = (ROOT / row['path']).read_bytes()
+        need(hashlib.sha256(raw).hexdigest() == row['sha256'], 'live source pin differs')
+        docs[row['path']] = json.loads(raw)
+    claims = json.loads((HERE / 'CLAIMS.json').read_text())
+    inspect_scope(docs, claims)
+    count = 0
+    # Change both the evidence transcription and decoded source: the scope
+    # anchors still reject a matching but materially broadened contract.
+    for index, before, after in [
+        (0, 'No broader allocation-equivalence proof or new authority framework is required.',
+         'A broader allocation-equivalence proof is required.'),
+        (1, 'not a universal allocation-equivalence proof', 'a universal allocation-equivalence proof'),
+        (2, 'do not require construction', 'require construction'),
+    ]:
+        bad_docs, bad_claims = copy.deepcopy(docs), copy.deepcopy(claims)
+        row = bad_claims['resolved'][index]
+        changed = row['source_text'].replace(before, after)
+        need(changed != row['source_text'], 'tamper placement failed')
+        row['source_text'] = changed
+        parts = row['pointer'].split('/')[1:]
+        parent = bad_docs[row['input']]
+        for key in parts[:-1]:
+            parent = parent[int(key)] if type(parent) is list else parent[key]
+        parent[parts[-1]] = changed
+        rejected(lambda: inspect_scope(bad_docs, bad_claims), str(index))
+        count += 1
+    for key in ('conditional_PR6347_witnesses_withdrawn', 'new_runtime_or_activity_acceptance'):
+        bad = copy.deepcopy(claims)
+        bad[key] = True
+        rejected(lambda: inspect_scope(docs, bad), key)
+        count += 1
+    bad = copy.deepcopy(claims)
+    bad['resolved'][0]['source_text'] += ' unbound transcription'
+    rejected(lambda: inspect_scope(docs, bad), 'transcription')
+    count += 1
+    print('PASS_G12_ORIGINAL_CONTRACT_SCOPE source_inputs=3 tamper_controls=' + str(count) +
+          ' target_execution=false original_acceptance=false')
+
+
+if __name__ == '__main__':
+    main()