Mega Gate — R2-D2 Channel Authority Recommendation

Date: 2026-06-18 · Workstream: LEGO-PILOT-SLICE-0-B2-MEGA-GATE-BUNDLE-2026-06-18 (Deliverable 4 of 20) · Editorial revision: rev1 Class: design-only / channel recommendation / decision-support · READ-ONLY · NON-ENACTING · NON-AUTHORIZING · NOT remediation · NOT technical design · NOT implementation · NO blocker resolved · NO runtime touched.

Metadata convention. Editorial revision (rev1) only. AgentData storage revision and content_length are authoritative in AgentData metadata at read time; not pinned in this body.

Channel-authority lock — the central lock of this packet. This packet states a RECOMMENDATION_ONLY — NOT AUTHORITY — OWNER_GATE_REQUIRED — FUTURE_TECHNICAL_DESIGN_REQUIRED preference among the candidate channels. It does not select, use, implement, wire, or promote any channel. No channel is made authority. The channel is a replaceable internal of B2 (B2-AC-7); whichever the Owner later chooses must stay inside the block so the contract is unaffected. If any wording here reads as a selection, that is CHANNEL_AUTHORITY_DRIFT and must be corrected to recommendation-only.

0. Status and non-authorization

STATUS: PASS — engineering / design-only. This is a complete design-only channel recommendation packet: the carried 5-channel classification, the recommendation-only preference among the two candidates, the explicit non-selection statement, the anti-overreach guardrails, and a Codex check-list for whether the recommendation overreaches.

Engineering PASS ≠ authority PASS. A PASS means the recommendation is complete and fair on paper. It is not an Owner authorization to select, wire, install, promote, or build any channel. Default disposition: HOLD.

Pipeline position (downstream-only). Deliverable 4 of the Mega Gate Bundle; it carries forward the accepted R2-D2 channel decision packet (Deliverable A of the planning bundle) as a recommendation for an eventual Owner decision. It selects nothing and opens no package.

Non-authorization (explicit). As Deliverable 1 §0, and specifically: it does not select or wire a channel; does not install pg_cron or any extension; does not promote any agent-api contract DRY_RUN→REAL_RUN; does not enable any queue worker/master switch; writes no scheduler/runner/cron spec. v0.1/FIX7 V3 not overwritten; v0.2 not authority.

Evidence basis — INHERITED_EVIDENCE. No runtime queried. Channel facts inherited from R2a + the accepted channel decision packet. AgentData metadata authoritative at read time. CAV-3/CAV-4/CAV-5 bound what is provable about each channel.

Reading discipline (Codex caveat, honored). All sources read directly from AgentData KB, bounded/sequential, by the main process — no parallel/background reader-agents, no sub-agents, no local-prose inference. /tmp = decode-scratch only, never SSOT.

1. Purpose

Answer the macro's third question: which channel should be recommended for an Owner decision, without becoming authority? The packet answers:

What is the carried 5-channel classification? — §3.
What is the recommendation-only preference, and why? — §5.
What makes this a recommendation and not a selection? — §5 + §0 lock.
What would constitute overreach, and how is it guarded? — §5 + §7.
What must Codex check? — §8.

The one rule, above all detail. The channel is a replaceable internal, not the block boundary. B2's contract (read uncertified → write inspect_* only, fail-closed) is channel-independent. This packet may recommend; it may not select, and it may not let any channel leak into the contract.

2. Sources read

All 25 required sources read first-hand from AgentData KB, by the main process, sequentially; none SOURCE_NOT_READ (full list in Deliverable 20 §2). Used principally: the accepted R2-D2 channel decision packet (Deliverable A — the 5-channel matrix, dispositions, recommendation-only posture, B2-AC-7); R2a root-cause (the channel substrate evidence); the B2 TD-prep §11 channel matrix + its Codex review ("channel matrix conceptual only; no channel selected as authority"); Điều 32 (DOT-100% / no manual SQL / no curl bypass); operating-rules (Assembly First).

3. Accepted baseline (carried, not re-derived)

The carried 5-channel classification (from the accepted channel decision packet, Deliverable A §15; dispositions unchanged):

Channel	Carried disposition	Carried reason (substance)
host cron	candidate	lowest blast radius; the proven channel for the sibling scanner DOTs; matches the inspector's periodic-scan nature; best LEGO replaceability; opaque to in-DB observability (visible only via the DB-captured `wf_host_crontab_snapshot`); not transactional with the `inspect_*` writes
agent-api executor (`:8090`)	candidate	existing healthy runner (Up 13 days); contract-bound and most observable / Điều-32-promotable; gated by master switches OFF + a `DRY_RUN→REAL_RUN` promotion; shared infra with the KG lane (must not couple the lanes)
pg_cron	risky / future-gated	clean in-DB atomicity but a net-new extension install (not installed today) → higher infra blast radius
job_queue worker	risky / future-gated	decoupled, but the substrate is disabled/idle and the undrained-queue failure mode (`event_outbox` ~215k undrained) is already present
manual one-shot	reject as standing channel	not testable/replaceable/rollbackable as a block; violates Điều 32 §2.1 (no manual SQL / no curl bypass); is the 2026-03-21 fused-INSERT vehicle; the only legitimate residual one-shot is B5, a separate Owner-gated backlog pass

Carried runtime substrate (INHERITED_EVIDENCE, R2a): master switches OFF (execute_enabled/real_run_enabled=false, dry_run_only=true; queue.worker.enabled=false; queue.job_substrate.enabled=false); pg_cron NOT installed; queue idle since 2026-05-26; no host-cron birth entry (the 0 6 slot is dot-nrm-lifecycle); the inspector DOTs DOT-TAC-BIRTH-VERIFY/-GATE are unwired stubs; the agent-api executor serves only the KG EXPLAIN DRY_RUN pilot today.

Carried Codex acceptance: "The channel matrix is conceptual only. … No channel is selected as authority." (B2 TD-prep + planning-bundle Codex reviews.)

4. Analysis — recommendation vs. selection

A recommendation ranks the candidates and explains the trade-offs for an Owner who will decide; a selection would name one channel as the channel B2 uses. The line between them is the channel-authority lock. This packet stays on the recommendation side by:

ranking only the two candidate channels (host cron, agent-api executor) and explaining their trade-off (Deliverable 5 deepens this), while leaving the choice explicitly to the Owner;
not naming a channel as "the" channel, not writing any wiring/contract/install spec, and not implying any default that bypasses the Owner;
keeping pg_cron / job_queue as risky/future-gated and manual one-shot as rejected, unchanged from the carried matrix;
restating B2-AC-7: whichever channel is later chosen stays inside B2; the contract does not change with the channel.

The recommendation is information for a decision, not the decision.

5. Channel recommendation (recommendation-only)

RECOMMENDATION_ONLY — NOT AUTHORITY — OWNER_GATE_REQUIRED — FUTURE_TECHNICAL_DESIGN_REQUIRED.

Channel	Disposition (carried)	Authority status
host cron	candidate — lowest blast radius; proven for the sibling scanner DOTs; best LEGO replaceability	Not selected
agent-api executor (`:8090`)	candidate — existing healthy runner; most observable / Điều-32-promotable; gated by master switch + `DRY_RUN→REAL_RUN` promotion	Not selected
pg_cron	risky / future-gated — net-new extension install (not installed today)	Not selected
job_queue worker	risky / future-gated — idle/undrained substrate; undrained-queue failure mode already present	Not selected
manual one-shot	reject as standing channel — Điều 32 §2.1 violation; fused-INSERT vehicle; only legitimate residual one-shot is B5 (separate)	Rejected as standing channel

The recommendation, stated as a preference for an Owner decision (not a selection):

If the Owner later opens B2's technical design, the two viable standing candidates are host cron and the agent-api executor. These are the two candidate channels; pg_cron and job_queue worker are risky/future-gated; manual one-shot is rejected as a standing channel.
Expectation (recommendation-only): host cron may be the fastest / lowest-blast-radius option — a single crontab entry outside the block, the proven channel for the sibling scanner DOTs, matching the inspector's periodic-scan nature, with the smallest swap cost.
Expectation (recommendation-only): the agent-api executor may be the more governed / observable option — contract-bound (in-DB), Điều-32-promotable, with dispatch records — at the cost of being gated by the master switches and a DRY_RUN→REAL_RUN promotion, and of sharing infra with the KG lane (which must not couple the lanes).
No channel is selected, used, wired, installed, promoted, or built. The choice between the two candidates (and the trade-off between speed/blast-radius and governance/observability) is an Owner decision, deepened in Deliverable 5 and gated by the proof-obligations in Deliverable 6. Whichever is chosen stays inside B2 (B2-AC-7).

Forbidden wording check (self-applied). This packet uses none of: "Selected channel:", "Use this channel:", "Implement this channel:", "Wire this channel:", "Promote this channel:". It uses only the recommendation-only formula above. Any drift to a selection is CHANNEL_AUTHORITY_DRIFT → HOLD.

6. Owner-gated future work

Every action below is forbidden now (OWNER_GATE_REQUIRED). Listing is scoping, not authorization.

Future work	Gate required	Forbidden now?
Select the B2 invocation channel (Owner's act, informed by this recommendation)	Owner decision	Yes
Wire host cron for the inspector	Điều 32 + S2 owner	Yes
Bind/promote the agent-api contract (`DRY_RUN→REAL_RUN`)	Điều 32 + S2 + contract promotion	Yes
Install pg_cron (extension)	Điều 32 + infra/extension approval	Yes
Enable the job_queue worker / master switch	Điều 32 + master-switch flip	Yes
Prove the chosen channel's liveness	the Deliverable 6 proof-obligations (read-only where possible)	Yes (not done here)

7. What remains unresolved

CHANNEL not selected (by design). This packet recommends; the Owner decides. CHANNEL_AUTHORITY_DRIFT guarded.
Substrate currently fail-closed. Switches OFF, queue idle, no birth cron, pg_cron absent — every standing channel that uses the runner/queue substrate is disabled and would require an Owner-gated enable.
CAV-3 / CAV-4 / CAV-5 carried. Channel liveness is provable only via DB-captured snapshots (no crontab -l/systemctl/docker exec tool; /opt/incomex/dot/bin + env unreadable); the transient GUC layer is unreadable. No overclaim is made about any channel's live state.
Overreach risk noted. The single overreach risk in a recommendation like this is letting "the two candidates are host cron and agent-api" harden into "use host cron." This packet guards it by stating both as candidates, recommending the trade-off be decided by the Owner, and writing no wiring/spec — but Codex must verify (Deliverable 19 / §8).
Blockers — all OPEN, none resolved: CONS-002, CONS-003, CELL-003/004/007, HOLD-1, HOLD-2, RISK-BYPASS, GOV-016/017, GOV-REUSE-001, Điều 39 runtime-EMPTY, Điều 35 production-readiness FAIL.
FUTURE_TECHNICAL_DESIGN_REQUIRED (NOT written here): any scheduler/runner/cron implementation, contract shape, dispatch payload, extension-install plan, worker-enable plan, or command sequence.

8. Ready for GPT/Codex review

Yes — as a design-only channel recommendation, not a selection.

Codex must check (anti-overreach):

Is the recommendation stated as RECOMMENDATION_ONLY — NOT AUTHORITY throughout, with no channel marked "selected"/"use"/"wire"/"promote"?
Are both host cron and agent-api kept as candidates (no premature collapse to one)?
Are pg_cron / job_queue kept risky/future-gated and manual one-shot rejected, unchanged?
Is B2-AC-7 (channel = replaceable internal) restated, with no channel leaking into the contract?
Is any scheduler/runner/cron/contract/install spec present? (Must be none.)
Does the recommendation overreach by implying a default that bypasses the Owner? (Must be no.)

Core rule, kept above all detail: the two candidate channels (host cron, agent-api executor) are recommended for an Owner decision; the speed/blast-radius vs. governance/observability trade-off is the Owner's to resolve; no channel is selected as authority, and whichever is chosen stays inside B2.

Default disposition: HOLD. Engineering PASS = a complete, fair recommendation on paper; it is not an Owner authorization to select, wire, install, promote, or build a channel. No PASS authorizes writes. All blockers remain OPEN.