<!-- DOC_STATUS: ACTIVE_NON_AUTHORITY -->

FIX7 Authority / N-Node — TKT v0.2-Aligned Review Packet (2026-06-11)

Host: T1 / Claude Code. Status established: FIX7_AUTHORITY_N_NODE_TKT_V02_ALIGNMENT_NEEDS_OWNER_OR_CODEX_RATIFICATION. Authority of everything in this packet: NON_AUTHORITY / NOT_PROMOTED / NO_SEAL. This packet is a request to review. It seals nothing, approves nothing, promotes nothing, and creates no real N7/N8/P7.

Codex NOT consulted. Owner approval NOT requested. Production / PG / Directus / registry / system_issues NOT mutated. T2 v0.2 framework NOT mutated. FIX7/V3 baseline NOT mutated.

---

0. One-paragraph summary

The FIX7 authority-seal lane is re-checked against the actual governed files using the T2 v0.2 proof convention (this is a consumption of T2's framework, not a parallel framework). The encoder was byte-exact reconstructed from KB (sha 13344f92…, matching both the governed packet pin and the N-node model JSON's S3.encoder_sha256). Three independent fail-closed harnesses run green on fresh reconstruction; tamper proofs confirm the integrity gate is not fake-green. The N1/N2/N9 numbering drift (MC-1/MC-2) is reconciled into a machine-readable canonical convention with an executable checker — but that canonical numbering is labels-only and requires owner/Codex ratification to be binding, so the final status is needs ratification, not Codex-ready.

1. Lane boundary (consumed, not invented)

• T2 owns the Tool-Kiem-Thu v0.2 proof-framework conventions (review-packet structure, commands.sh/exit_codes.json/HASH_MANIFEST/packet_tree, fresh reconstruction, report-vs-file audit, no-vector evidence escrow, authority firewall, second-pass adversarial probes, registry/index discipline). v0.2 remains NON_AUTHORITY / NOT_PROMOTED.
• T1 owns the FIX7 semantic / N-node / authority-seal lane. T1 consumes the T2 conventions above and does not mutate the v0.2 dev packet. No T2-framework gap was found that blocks this lane (see cross-lane alignment report).

2. What this packet PROVES (reproduced here, real exit codes)

Claim	Gate	Result
Encoder is byte-exact to the governed pin	G0_encoder_frozen	sha 13344f92… == governed pin (2 independent sources)
Encoder fails closed across 48 vectors	G1_encoder_selftest	48/48 PASS, exit 0
Codex's 8 direct fail-open probes + 11 more are now rejected	G2_codex_probes	19/19 REJECTED, 0 fail-open, exit 0
N-node numbering/label drift fails closed outside alias context	G3_n_node_checker	15/15 PASS (6 positive + 9 bad-label rejects), exit 0
Supplemental independent adversarial probes fail closed	G4_supplemental	10/10 fail-closed, exit 0
Packet integrity gate is fail-closed (not fake-green)	tamper proofs	file-tamper→FAIL, tree-tamper→FAIL, control→PASS

Reality headline: no invalid input accepted; no PASS/digest/seal emitted for any invalid input; no candidate/rehearsal promoted to authority; no real seal claimed.

3. What this packet does NOT prove / does NOT do

• It does not create a real N7/N8/P7 seal. The real path stays gated SEAL_REAL_N6_NOT_AVAILABLE.
• It does not ratify the canonical N-numbering. The checker encodes the model's proposed_correction_not_adopted as an engineering convention; binding ratification is owner/Codex (G-DOC-1).
• It does not unblock FIX7 implementation, run REAL_RUN/QT001/permit, or touch production.

4. How to reproduce

# rev2 (2026-06-11): all 5 manifest-listed modules are now PUBLISHED at this
# packet path (Codex CODEX-ALIGNMENT-PACKET-INCOMPLETE closed). authority_seal_encoder.py
# and codex_probes.py were published byte-exact to their governed pins
# (13344f92… / 112b4ec5… == fix7-authority-closure-2026-06-10/). No pre-fetch needed.
bash RERUN.sh        # step1 HASH_MANIFEST verify, step2 packet_tree, step3 commands.sh
shasum -a 256 -c HASH_MANIFEST.txt

Expected: RERUN_RESULT: PASS, OVERALL_RESULT: PASS, every gate OK.

5. Module provenance (no divergent duplicate authority)

File	sha256	source
authority_seal_encoder.py	13344f92…	published here byte-exact to fix7-authority-closure-2026-06-10/ rev2 pin (CODEX-ALIGNMENT-PACKET-INCOMPLETE closed)
codex_probes.py	112b4ec5…	published here byte-exact to fix7-authority-closure-2026-06-10/ rev1 pin (CODEX-ALIGNMENT-PACKET-INCOMPLETE closed)
n_node_checker.py	4fdf7d24…	new this lane (model-grounded N-node convention checker)
sp_align_probes.py	4b2e04e2…	new this lane (supplemental independent probes)
commands.sh	3e00d0b6…	new this lane (T2-convention gate ledger)

packet_tree (sha256 over HASH_MANIFEST.txt): 96d00b9e….

6. Blocker summary (all authority/owner — none are engineering defects)

Blocker	Kind	Owner	Blocks
N-number ratification (G-DOC-1/2/3)	authority/doc	owner/Codex	binding canonical N1/N2/N9 numbering
SEAL_REAL_N6_NOT_AVAILABLE (G-AUTH-1)	authority/provenance	owner/operator + Codex	real N6 corpus chain
G-AUTH-2 (A1/A2/A3/A5 + N8 signer + P7 ids)	authority	owner/Codex	real N7/N8/P7 inputs
G-OWNER-1 (OWN-1)	owner	owner	standing do-not-approve

7. Next decision (NOT decided here)

1. Owner/Codex ratify the canonical N-number→value table (cheap, no production effect) — then the checker convention becomes binding.
2. Only after ratification, run the real-N6 provenance macro (engineering half is T1-doable; authority half is owner/operator+Codex).
3. Authority seal N7→N8→P7 remains a Codex/owner act, still gated.

---

Self-Codex, T1/Claude Code. Codex NOT consulted. Owner approval NOT requested. v0.2 NON_AUTHORITY/NOT_PROMOTED and not mutated. No real seal. No production mutation.