CommHub#730/#1138 authenticated Dashboard queue deadline first slice.
Final source/core/public facts are recorded in report-test1138-human-queue-public.txt; the preparation and intermediate failures below are retained.
Baseline main07994c0813efdb4e8f192131e53753cd1b114e07 uses30m for every Codex FIFO request. New policy uses4h only for the existing authenticated Dashboard classifier, carried by a separate explicit runtime audience marker; generic steer permission remains insufficient. Explicit option/env precedes both defaults, execution/startup/Hub expiry/cancel checks remain separate. No insertion/active-CommHub-turn steering or queue-position UI is delivered here.

Validation is isolated Docker: existing origin/legacy/dedup classifier tests, fake-clock real runtime30m/4h/default overrides, wait-then-start execution budget and finite cancellation; actual transpiled production CLI function verifies marker propagation from the authenticated inbox classification. A container-only mutation disables that production marker and must fail the same assertion. Type and exact package-pair gates run in Actions. This report is preparation; final CI/main/source/public byte facts will be appended after completion. No current pass or release is claimed before the gates finish.

Candidate CLI175/Node141; notes split Install/Upgrade and retain original lifecycle authority/normal-idle adoption/rollback/state backup boundaries. Source only; no production/account/model/provider/config/schema/port change, real node restart, shared-thread switch, globalnpm install or credential read/copy. Actual adoption and original-user queue behavior remain pending; Hub expires_at can cut off even a protected queue window. No source-clone-only private-state disaster recovery promise.

Local validation budget preserved: initialDocker build reached90s cap while exporting/unpacking its dependency layer, no functional pass claimed. Cached-image run capped20s yielded70normalPASS/1fixtureFAIL (production CLI probe lacked the reportCapacityRetry dependency stub); timeout124 also retained. Added that controlled stub and procps for the existing isolated process-cleanup regression. Final exact-source Actions result, not this partial local run, is the merge gate.

Firstcandidate12be6bede3aece876d5378d70cabb73a94057cab dedicatedDocker38385059601/job115254673805 SUCCESS, testedmergea6885f8fed6f8d85c637bc544e1809f8f6a7fe16:71normalPASS/200assertions and1productionCLI mutationASSERTION RED. Pair/type/task-activity and10required contexts also passed. Documentation's existing30m generic statement has now been updated to the authenticated4h/other30m policy; this documentation/workflow/report follow-up does not change runtime code, and its final-source gates remain pending until Actions completes. No merge/release is claimed yet.

Follow-up4b12836f3 Docker38385818418/type/pair/activity PASS, but docs-integrity38385818438/job115257222247 FAIL: agent-node guide now contains a new preview-version behavior assertion and was not registered in the RELEASE-SOP inventory. Added the guide's lower-bound queue policy to that authoritative table, without weakening or bypassing the checker. Original failed run retained; final-source checks must pass before merge.

Final corrected CLI175 / Node141 pairing and anonymous public byte/SRI receipt are now recorded in report-test1138-human-queue-public.txt. Original intermediate failures remain above; native adoption and mother730 steering/UI remain open.
