KB-307C

TKT Base Scope and Output-Level Policy Report

5 min read Revision 1
<!-- DOC_STATUS: ACTIVE_NON_AUTHORITY -->

TKT Base — Scope & Output-Level Policy Report

Macro: TKT_BASE_STRUCTURAL_EVIDENCE_GOVERNANCE_PACKAGING_MACRO_2026_06_11 Status contribution: Workstream A. Authority: NON_AUTHORITY / NOT_PROMOTED. Date: 2026-06-11 · Host lane: T2 · Codex consulted: NO

This report defines the official TKT Base scope and the only output levels a base run may emit. The normative copies live in the pack: base/tkt-base-structural-evidence-governance-pack-2026-06-11/TKT_BASE_SCOPE_AND_LIMITS.md and …/TKT_OUTPUT_LEVEL_POLICY.md.


1. Scope statement

TKT Base is the structural-evidence and governance testing layer for technical packets. It is the honest floor: a packet that passes L0–L3 is structurally sound, reproducible, fail-closed, and governance-clean. It is not the semantic Text-as-Code / IU Smart-Brick layer and emits no semantic claim.

2. Output levels (official)

Level Name Pass condition (summary)
L0 FILE PASS every load-bearing file exists + hash matches; no missing file; sha256(manifest)==tree_pin
L1 PACKET RECONSTRUCTION PASS manifest + tree pass; commands/RERUN pass; exit_codes.json byte-stable
L2 FAIL-CLOSED PASS bad input rejected; invalid emits no PASS/cert/digest/seal; any_fail_open=false
L3 GOVERNANCE CONSISTENCY PASS IDs no-collision; no orphan; lanes respected; firewall holds; no-vector evidence has hash+pointer+regen

Levels are cumulative and ordered: L(n) is claimable only if L0..L(n) all pass. A FAIL at L(n) caps level_reached at L(n-1); higher levels report N/A, never PASS.

3. Forbidden overclaims (the ceiling)

A base run MUST NOT emit IU_TRACEABILITY_PASS (L4), SEMANTIC_TEXT_AS_CODE_PASS (L5), or RELEASE_BUNDLE_PASS (L6), and MUST NOT claim "the Tool is complete for semantic Text-as-Code." Any attempt is a fail-closed condition (harness probe P10, proven nonzero with no PASS token). The highest honest base verdict is L3.


4. Readback Table A — TKT Base capability

capability current artifact proof status reusable? gap level
file presence + hash checkers/manifest_file_presence_policy.md + harness P1/P2 PASS (10/10 probes) yes none L0
tree pin integrity checkers/packet_tree_policy.md; pack tree 6a6137d1… PASS (shasum -c clean) yes none L0
packet reconstruction RERUN.sh; RERUN_RESULT: PASS in mktemp PASS yes none L1
deterministic verdict exit_codes.json byte-stable 21e03bcb… PASS (byte-exact) yes none L1
fail-closed on bad input harness/tkt_base_harness.py 10/10, any_fail_open=false PASS yes none L2
report-vs-file audit checkers/report_vs_file_audit_policy.md + harness P5 PASS yes none L0/L3
authority firewall checkers/authority_firewall_policy.md (F1–F9) PASS (policy + P6) yes runtime probe is policy-level here L3
object-ID collision checkers/object_id_collision_policy.md + harness P7 PASS yes none L3
no-vector evidence checkers/nvsz_no_vector_evidence_policy.md + harness P8/P9 PASS yes real NVSZ root = operator blocker L3
overclaim guard TKT_OUTPUT_LEVEL_POLICY.md + harness P10 PASS yes none L2/L3

5. Readback Table B — TKT semantic-deferred

capability why deferred required IU input future level dependency
IU traceability no machine-readable IU relation graph a checker can consume IU ids + relation objects + metadata binding + traversal checker L4
semantic Text-as-Code no Text-as-Code grammar/linter/oracle yet grammar/schema + controlled vocab + semantic linter + committed semantic oracle L5
release/bundle depends on L0–L5 + release policy passing L0–L5 + release policy + bundle manifest + authority L6

Grounding references (read-only, not implemented): …/architecture/gpt-handoff-iu-text-as-code-smart-brick-foundation-gap-and-next-requirements-2026-06-01.md, …/dieu38-trien-khai/reports/gpt-analysis-information-unit-infrastructure-next-2026-05-01.md, …/dieu38-trien-khai/addendum/metadata-song-mieng-thong-tin-alignment-2026-05-01.md.

6. Verdict

Workstream A complete: scope defined, four output levels formalized, overclaim ceiling set at L3, semantic levels explicitly deferred. Evidence: the pack files above plus the validation run recorded in tkt-base-validation-probes-2026-06-11.md.

Back to Knowledge Hub knowledge/dev/laws/tool-kiem-thu/base/reports/tkt-base-scope-and-output-level-policy-2026-06-11.md