CommHub #730 / child #1141 authenticated human waiting priority — final source and public receipt.

Source/core:
PR2704 final candidate ecc1447ef605efca52f864af10b11851a98f8e94, tested merge 8bc695728266391bd3e73aca944b3304909b4c99, merged main 83cd8fc1cbc33be74832cb39632789cb438d1aaf at 2026-10-13T11:44:11Z. All three full Git trees equal 793fb7a20e029c87209b892d097b613fd0042477.
Exact-source Docker 38391860764/job115277345205:138 normal PASS,505 assertions,0 FAIL, plus four independent production ASSERTION RED witnesses (existing CLI audience, new CLI dispatcher classifier, runtime initial/retry marker, bridge priority gate). Type ratchet38391861042, pair38391861073, exact task activity38391861097, docs38391861209, mutation pins38391860893 and all10 required contexts passed. Full optional QA38391861111 and Docker E2E38391860847 were canceled only on this own merged PR after these gates; canceled does not mean PASS. Main/release/peer workflows were not canceled.

Intermediate failures retained:
First candidate15c2ac0e2 mutation-pin38391402013/job115275813381 FAIL: previous test584 concurrent-start mutation no longer matched the changed queue method, and new dynamic inline anchors increased the unchecked baseline. Re-pointed the old mutation at the equivalent current busy guard/enqueue and rewrote new targets/anchors as literal expressions; no checker baseline, assertion or production concurrency guard was weakened.
A supplemental local witness incorrectly required inherited584's failure to print expect(received); it reported fixture FAIL. A separate shell-quoting attempt failed before testing. Neither counts as product PASS. Final targeted Docker584 mutation reached its original semantic failure: submitTask of a second active task was incorrectly allowed through to low-level startTaskTurn, which rejected it; the named queue-behavior test failed (SEMANTIC RED). New four wiring witnesses independently reach actual behavior assertions. Original failure logs remain private; the tests use no real user conversations or credentials. Total local functional validation remained below two minutes; exact-manifest full core Actions is the main acceptance gate.

Behavior/limits:
Existing Hub-authenticated Dashboard classification drives an explicit audience marker across both inbox admission and runtime initial/capacity-retry bridge submission. Human rows still waiting lead machines, with FIFO within each class. Node/scheduler/legacy/alias-only and generic steer options alone do not promote work. Deferred drain restores its prior position within its own class. Existing authoritative Hub row checks, cancellation and exact task/turn reply attribution remain.
Running handlers/turns, an in-flight start or a row already claimed for a Hub check are not preempted. Saturated dispatcher waits for a slot; continued human arrivals can postpone machines. Runtime defaults remain4h authenticated Dashboard/30m other, explicit options/env and Hub expiry/cancel/permissions remain authoritative. Existing external human-TUI steering is retained; active-owned-CommHub-turn injection, queue-position UI, original-device confirmation and runtime adoption remain separate open work.

Recovery:
Release-v2.5.0-preview.142.md and release-v2.3.0-preview.176.md contain paired private-prefix Install and original-owner normal-idle Upgrade/actual-entrypoint/natural-task verification, prior-pair rollback and private-state backup boundaries. Startup/topology/secure secret sources remain in the existing operation/deployment runbooks; no production schema, ports, provider/model, supervisor configuration, login, node identity/cwd/original thread or running process was changed. Git restores software/procedures, not private login/history or production data. This release is not a fresh-machine disaster-recovery or native adoption pass.

Public verification:
Node142 and CLI176 must both pass all four exact-main release gates and anonymous metadata/full-tarball byte/SRI verification. Final facts are appended only after those observations.

CLI176 first formal attempt38393452382 gate1/job115282926682 FAILED after candidate version/help succeeded: registry-selected Hub138 remained at dependency resolution, /health did not respond2xx and startup was reported failed. Exact cause remains unknown; this is not proof of a network failure or new queue-source defect. Other three gates succeeded, publish SKIPPED. agent-network/bin/cli.ts is byte-identical to the passing CLI175 main1b85 source. Original attempt/job log retained; one bounded --failed rerun uses the same run/gated artifact and exact main83cd, while Node142 is not rebuilt or republished. No original-node lifecycle was touched and no gate was bypassed.
Attempt2 gate1/job115284808776 passed all installation smoke cases, with the other three gates green and the same sealed gated archive (2126993 bytes, sha25607042f6723a461e033c755b3cdac16c701793b15ec97dcec453d615f4e894a2c). Successful retry does not establish the first startup failure's root cause. Registry observation remains the final public proof below.


Final paired public observations:
Node142 official npm run 38392369685 all SUCCESS, build source 83cd8fc1cbc33be74832cb39632789cb438d1aaf, observed 2026-10-13T11:53:51.258559+00:00, 594714 bytes, SHA256 bf27ed629d770e1e91bc0e5b685d486bc54dc3e1479a392f3f0bdfafc6ace692, registry SRI sha512-ZXfy/0YSgqnRaV6IDR6vtZX566DceZIQDSCSktpVlLsy3F+Rw+Me2+n101KtfkxiFUZ4/Z6hs3aQSytUifqSjw==; anonymous public bytes match its CI gated archive. Tags {'latest': '2.5.0-preview.58', 'preview': '2.5.0-preview.142'}, latest matches baseline.
CLI176 official gated-download workflow 38395966269 SUCCESS, immutable tag cli-v2.3.0-preview.176 resolves exact main83cd. Anonymous full tarball observed 2026-10-13T12:19:22.727982+00:00, 2126993 bytes, SHA256 07042f6723a461e033c755b3cdac16c701793b15ec97dcec453d615f4e894a2c; public bytes match the sealed original CI archive. Calculated SHA512 SRI sha512-JT23bW/qnTwQ3K2cTOiZAwfKNwELfTBWWPjPMVML9vb9PmO6DX+T+1BWwdReg+HuW0JA2DxgK53Tz9GZMCcDDg== (not a verified npm registry integrity field). Official asset https://github.com/sleep2agi/agent-network/releases/download/cli-v2.3.0-preview.176/sleep2agi-agent-network-2.3.0-preview.176.tgz.
CLI176 original npm run38393452382 attempt2 ended FAILURE solely at the final registry recheck after all four gates and npm publish step SUCCESS. Both version endpoint and full package index still lacked176 at12:13:24Z (preview175/latest76). This failure is retained, not relabeled PASS; npm visibility/integrity remains a separate follow-up and no npm republish or retag was performed. Both public artifacts were built from exact merged main83cd and remain current-main ancestors; source/core/public download1141 is complete, original adoption/active-owned-turn injection/queue UI remain open.

Delayed npm confirmation:
Anonymous exact CLI176 version endpoint returned200 at2026-10-13T12:26:14Z. Full npm metadata/archive proof observed 2026-10-13T12:27:40.929078+00:00: 2126993 bytes, SHA256 07042f6723a461e033c755b3cdac16c701793b15ec97dcec453d615f4e894a2c, SHA1 d5c10ca63a499a792f09ab7501dece583ccbd1ad, registry integrity sha512-JT23bW/qnTwQ3K2cTOiZAwfKNwELfTBWWPjPMVML9vb9PmO6DX+T+1BWwdReg+HuW0JA2DxgK53Tz9GZMCcDDg==; npm public bytes equal the original gated artifact and official GitHub download. preview=176/latest=2.3.0-preview.76, latest unchanged. The main83cd producer and current-main ancestry remain verified. This closes the separate npm visibility/SRI follow-up1143 without republishing or retagging. Original installation failure and later registry-timeout workflow FAILURE remain historical failures; delayed public availability does not relabel them SUCCESS. The exact propagation/startup cause is still unknown.
