KB-6B9C

RS-TKT-0A · 00 Survey Source Inventory

13 min read Revision 1
tool-kiem-thulegolaws-newrs-tkt-0asurveysource-inventoryread-onlynon-authorizing2026-06-21

RS-TKT-0A · 00 — Survey Source Inventory

Lane: RS-TKT-0A — Tool-Kiem-Thu LEGO Survey & Conversion Plan Path: knowledge/dev/laws-new/tool-kiem-thu-lego/00-survey-source-inventory-2026-06-21.md Date: 2026-06-21 Verdict (rollup): RS_TKT_0A_READY_FOR_GPT_REVIEW (see 08-final-survey-report-…) Gate: REGISTRATION_HOLD · REGISTRATION_CAN_PROCEED = NO · 0 runtime mutations (KB design-doc writes only) Authority: NON_AUTHORITY · may_gate=false · decision_effect=NONE · survey/design-only


0. Method (how this inventory was built)

  • Reconstructed from actual governed files, not from memory or summary prose. Every path below was listed/read via the AgentData KB tools (list_documents, get_document, get_document_for_rewrite, batch_read, search_knowledge).
  • Five read-only survey passes were run (old TKT corpus; P6 checker/DOT; laws-new LEGO; RS5A/RS5B Codex pain points; NVSZ run-evidence). No upload/update/patch/delete of any source doc; no VPS write_file; no PostgreSQL writes.
  • Harness not trusted as proof. Counts marked "listed" come from list_documents pagination to next_offset=null. Where a count is approximate it is labelled.
  • This file is an index of evidence; the reuse decisions live in 01-…reuse-gap-map.

1. Source classification buckets

# Bucket Primary KB prefix Read depth
A Old Tool-Kiem-Thu knowledge/dev/laws/tool-kiem-thu/ full on base pack + contracts + scope reports; summary on checkpoints
B P6 Checker/DOT design knowledge/dev/laws/dieu38-trien-khai/ + knowledge/dev/reports/ full (4 docs)
C NVSZ / no-vector evidence TKT base pack …/checkers/ + …/v0.2-hardening/review/ full on policy, manifest, validator pack
D laws-new LEGO knowledge/dev/laws-new/ full on SSOT + new-iu + pilot; summary on F0–F5/FX
E RS5A/RS5B pain points knowledge/dev/laws-new/reports/rs5a*, rs5b/ full on index/rollup/decision/codex-packet
F Codex review examples knowledge/dev/laws-new/reports/codex/ full on the RS5A-chain verdict reviews

2. Bucket A — Old Tool-Kiem-Thu (knowledge/dev/laws/tool-kiem-thu/)

Total documents listed: 433 (paginated 100+100+100+100+33; next_offset=null confirmed). Approx bucket split: index/readme 2; base structural-evidence-governance-pack ~29 (the reusable prize); base-reports 4; checkpoints ~48; contracts 4; designs ~25; governance/object-registry ~20; planning ~13; reports ~70; reviews ~9; FIX7/recheck-9/v0.2 packets ~120; support ~10.

A.1 The reusable base pack (most load-bearing)

Prefix …/base/tkt-base-structural-evidence-governance-pack-2026-06-11/, tree pin 6a6137d1…:

Path (relative to base pack) Type Relevance
TKT_BASE_SCOPE_AND_LIMITS.md scope HIGH — L0–L6 model, "honest floor"
TKT_OUTPUT_LEVEL_POLICY.md policy HIGH — only verdicts TKT may emit; overclaim guard
checkers/manifest_file_presence_policy.md checker policy (L0) HIGH
checkers/packet_tree_policy.md checker policy (L0/L1) HIGH
checkers/report_vs_file_audit_policy.md checker policy (L0/L3) HIGH
checkers/fail_closed_probe_policy.md checker policy (L2) HIGH — P1–P10 + detector-correctness rule
checkers/authority_firewall_policy.md checker policy (L3) HIGH — F1–F9, no-seal
checkers/object_id_collision_policy.md checker policy (L3) MEDIUM
checkers/nvsz_no_vector_evidence_policy.md checker policy (L3) HIGH — see bucket C
limitations/TEXT_AS_CODE_SEMANTIC_DEFERRED.md boundary HIGH — L5 deferral
limitations/IU_INPUT_REQUIREMENTS_FOR_LEVEL_4_5_6.md boundary HIGH — L4–L6 inputs
manifest.json self-desc HIGH — NON_AUTHORITY + raw_evidence escrow block
exit_codes.json verdict anchor HIGH — G0–G3 gates
commands.sh, RERUN.sh self-verify runner HIGH
HASH_MANIFEST.txt, packet_tree.sha256 integrity HIGH
harness/tkt_base_harness.py probe harness HIGH (reference only — runtime, not ported)
packet_template/* copy-me skeleton HIGH
examples/{v02-review-packet,nvsz-dryrun,fix7-n6}-example.md examples MEDIUM

A.2 Base reports (…/base/reports/)

tkt-base-scope-and-output-level-policy-2026-06-11.md · tkt-base-semantic-text-as-code-deferred-boundary-2026-06-11.md · tkt-base-validation-probes-2026-06-11.md · tkt-base-packaging-report-2026-06-11.md. Relevance HIGH.

A.3 Contracts (…/contracts/)

authority-contract-v0-1-2026-06-09.{md,json} (v0.1 DOT-coupled authority boundary) · b7-governed-export-step-contract-2026-06-10.md · offline-mvp-governed-packet-consumption-contract-2026-06-10.md. Relevance: discipline reusable, surfaces stale.

A.4 Index / readme / designs / checkpoints

00-index.md (rev 117, living changelog) · README.md. Designs include article14-executable-evidence-5lens-detector-spec-2026-06-10.md (HIGH), acceptance-test-matrix-implementation-package-dot-v0-1-2026-06-09.md. Checkpoints capture FIX7 executable approval lane, Phase-2 offline MVP, B7 export packet, sandbox attestation (process history — archive, not port).


3. Bucket B — P6 Checker/DOT design

Path Type Status Relevance
knowledge/dev/laws/dieu38-trien-khai/P6-checker-dot-design-v0-2.md design (OFFICIAL v0.2) read full HIGH — checker contract thinking
knowledge/dev/reports/gpt-review-p6-checker-dot-design-v0-round1-2026-04-26.md review (22 findings) read full MEDIUM
knowledge/dev/reports/gpt-review-p6-checker-dot-design-v0-1-round2-2026-04-26.md review (12 findings) read full MEDIUM
knowledge/dev/reports/gpt-review-p6-checker-dot-design-v0-2-final-pass-2026-04-26.md review (final pass) read full MEDIUM

Note: intermediate v0 / v0.1 design revisions are not in KB; only final v0.2 survives. Final-pass review asked for a canonical upload at …/P6-checker-dot-design.md (no suffix); the artifact in KB carries the -v0-2 suffix (cosmetic drift, not blocking).


4. Bucket C — NVSZ / no-vector evidence

Path Type Relevance
…/base/…pack…/checkers/nvsz_no_vector_evidence_policy.md policy R0–R8 + exit taxonomy HIGH
…/base/…pack…/manifest.json (raw_evidence block) escrow records HIGH
…/base/…pack…/examples/nvsz-dryrun-example.md dry-run HIGH
…/dev/v0.2-hardening/review/v02-nvsz-root-provisioning-dryrun-2026-06-11/NO_VECTOR_POINTERS.md live pointer table MEDIUM
…/v02-nvsz-root-provisioning-dryrun-2026-06-11/nvsz_root_requirements.json machine R0–R8 (TKT-OBJ-291) MEDIUM
…/designs/v02-evidence-escrow-model-2026-06-11.md; validators nvsz_root_validator.py, nvsz_escrow_validator.py model + validators reference only (runtime)

5. Bucket D — laws-new LEGO (knowledge/dev/laws-new/)

Path Type Relevance
tool-kiem-thu-lego/index.md new workspace index HIGH (quoted verbatim in 02)
README.md track overview HIGH
matrix-refactor-implementation-plan.md SSOT (io_contract.v0.1) HIGH
matrix-refactor-quick-rules.md SSOT (R-rules, S-rules) HIGH
matrix-stamp-governance-addendum.md SSOT (stamp lifecycle) HIGH
de-bai-cai-tien.md problem statement (Lego Protocol) HIGH
new-iu/de-bai-mieng-thong-tin-moi-lego.md New IU direction (anti-island rules) HIGH
new-iu/dat-van-de-sua-iu.md New IU problem statement MEDIUM
newlaws/consolidation/lego-pilot-slice-0-staging-io-contract-2026-06-18.md staging IO contract HIGH
newlaws/consolidation/lego-pilot-slice-0-bad-input-delete-fast-verification-plan-2026-06-18.md bad-input plan (BAD-1..15) HIGH
f0-…,f1-…,f2-…,f3-…,f4-…,f5-…,fx-…reuse-survey-packet.md module/contract dimensions MEDIUM (F4 highest — checker/promote/rollback)

6. Bucket E — RS5A/RS5B packets (knowledge/dev/laws-new/reports/)

Standard packet file-set observed = index + contiguous 01..NN (last = decision-packet) + codex-review-packet, inside reports/<stage>/, plus one top-level rollup reports/macro-<stage>-….md.

Directory Count Notes
reports/rs5a/ 14 original G2 dossier; 09-negative-authority-test-suite-80-cases-…md filename says "80" but body = 84 (drift)
reports/rs5a-patch3/ 7 lifecycle / quorum-precedence / replay-G08
reports/rs5a-patch4/ 7 total Q-order / eval unit / compound fixtures / G02 domain
reports/rs5b/ 9 G2 owner-of-record execution/authorization design (non-mutating)
reports/ (top level) rollups macro-rs5a…, macro-rs5a-patch1..4…, macro-rs5b…

7. Bucket F — Codex review examples (knowledge/dev/laws-new/reports/codex/, ~41 docs)

The RS5-chain verdict-bearing reviews (chronological), used as the adversarial template for 06:

Path (under reports/codex/) Verdict
codex-review-rs5a-g2-owner-of-record-decision-2026-06-21.md NEED_PATCH1
codex-review-rs5a-patch1-prerequisite-sequencing-correction-2026-06-21.md NEED_RS5A_PATCH2
codex-review-rs5a-patch2-semantic-closure-precision-2026-06-21.md NEED_RS5A_PATCH3
codex-review-rs5a-patch3-deterministic-lifecycle-and-oracle-predicates-2026-06-21.md REJECT_RS5A_PATCH3_QUORUM_PRECEDENCE_INSUFFICIENT
codex-review-rs5a-patch4-total-quorum-precedence-and-g02-domain-2026-06-21.md ACCEPT_RS5A_PATCH4

RS5B has no Codex review yet (its codex-review-packet exists; it goes to GPT first).


8. MISSING_CONTEXT_BLOCKER

ID Missing item Why it matters Impact if unresolved Safe fallback Recommended next action
MCB-1 RS5B has no external Codex review (only self-check BI01–BI10) The RS5B defect classes are self-reported, not externally caught 06 RS5B checks are partly inferred, not yet Codex-validated Treat RS5B checkers as DRAFT; mark RS5B-specific rows "self-reported" Run 06 profile against RS5B once Codex reviews it
MCB-2 Two NVSZ exit taxonomies (escrow validator vs root-provisioning validator) with non-identical numeric maps (e.g. invented-root = escrow 9 / root-validator 4) A run-evidence checker must gate on exactly one taxonomy Ambiguous failure codes Pin the escrow taxonomy as canonical in 05; document the divergence Owner/Codex to confirm canonical taxonomy
MCB-3 Ledger filename inconsistency: base pack ships HASH_MANIFEST.txt; R6 run-packet layout names it HASH_MANIFEST.sha256 Affects mechanical file-presence checks False FAIL on a correct packet 05 records both; checker accepts either, warns on mismatch Reconcile to one name at design approval
MCB-4 rs5a/09 filename "80-cases" vs in-doc "84 cases" Filename↔title drift is itself a checker class Confusion on count Treat body (84) as authoritative; flag filename Rename on next additive pass (do not edit prior file)
MCB-5 No designated NON_VECTOR_ROOT (blocker V02-PB-NVSZ-1, owner/operator only) TKT run evidence cannot be escrowed to a real root yet L1/L2 run evidence stays local-hashed-regenerable (honest interim) Keep nvsz_root.designated=false; never invent root Owner/operator designation (out of this lane's authority)
MCB-6 No single authoritative "laws-new architecture" doc; direction inferred from 3 SSOT files + new-iu + pilot Conversion plan rests on inference Risk of mis-stating direction Cite the 3 SSOT files as binding; mark inferences GPT/Owner confirm the SSOT set is current

9. Read-status summary

  • Read in full: base pack (scope, output policy, 7 checker policies, 2 limitations, manifest, exit_codes, commands.sh, RERUN.sh, HASH_MANIFEST, packet_tree, examples); base reports; v0.1 authority + offline-MVP contracts; P6 v0.2 + 3 GPT reviews; NVSZ policy + manifest + dry-run + root requirements; laws-new 3 SSOT + README + de-bai + new-iu pair + pilot pair; RS5A-patch3/patch4 index+topics+decision+codex-packet; RS5B index+topics; 5 RS5A-chain Codex reviews; new workspace index.
  • Read at summary level: TKT checkpoints (~48), FIX7/recheck-9 packets (~120), F3/F5/FX reuse packets, older RS1–RS4 reviews.
  • Not found: P6 v0/v0.1 intermediate revisions; a standalone V02-PB-NVSZ-1 blocker doc (carried inline in R8 + dry-run).

This inventory satisfies completion-proof items (1) "actual list of files read." No source mutation performed.