The production lane is the `.NET` cloud in `src/Jibo.Cloud/dotnet`. The Node server remains the protocol oracle, capture harness, and fast reverse-engineering lab, but it is no longer the long-term hosted architecture.
Day-to-day feature sequencing lives in [feature-backlog.md](feature-backlog.md). Live closeout checks live in [regression-test-plan.md](regression-test-plan.md). This file tracks release shape, current code truth, evidence sources, and the boundary between `1.0.18` closeout work and `1.0.19` follow-up work.
Release `1.0.18` is now in feature-hardening. Its main bug-fix theme is alarm and photo/gallery behavior on stock OS `1.9`, with a few small feature slices added while the test loop is warm.
- Before the cloud-version test, the robot's local `jibo-server-service` restarted after a broken pipe, then `ssm` raised `Q4-Server_connection_lost` and local `@be/settings` opened the connection-lost error path. The notification connection recovered about 31 seconds later. Treat early-test confusion as suspect if this local-server recovery appears in the same window.
- The cloud-version answer itself proved the running build was `1.0.18`, but the previous source treated `cloud_version` as a follow-up conversation. A fresh hotphrase `LISTEN` then captured speech tail as `Cloudford.`, and generic chat replied `thanks. I heard, Cloudford.`
- Current source now makes `cloud_version` a one-shot diagnostic, uses a longer diagnostic speech-tail ignore window, and ignores no-transcript hotphrase launch `LISTEN` setup packets inside that window. The existing no-`LISTEN` binary guard already ignored same-transID binary tails after finalization, but Test 27 showed it could not stop a brand-new hotphrase listen by itself.
- Test 28 showed our cloud-version/generic Nimbus `LISTEN` match entering stock BE with `skipSurprises` unset. After Nimbus settled, BE requested local `@be/surprises`; Test 28 inhibited the offer because VAD heard people talking, while Test 27 used the same doorway to select `@be/surprises-ota` and speak the backup-in-progress warning.
- Current source now emits `match.skipSurprises = true` for hosted turn results, fallback matches, and local skill redirects. Stock BE maps that to `skipSurprisesExternal`, preventing normal cloud replies from falling into end-of-skill surprises such as OTA/backup prompts.
- Test 29 showed the deployed `skipSurprises` payload in the robot logs and did not produce another backup announcement in the focused run. It still interrupted cloud-version speech because the spoken phrase `Open Jibo Cloud version...` included `Jibo`; stock Nimbus runs the response as a runtime MIM, and the local hotphrase detector stopped TTS before our cloud-side late-listen ignore could help.
- Current source now speaks the diagnostic as `Cloud version ...` without saying `Jibo`, while keeping the one-shot and late-listen cleanup guards.
- Backup-in-progress still appears robot-local. Tests 27, 28, and 29 had no matching `Backup_*` HTTP calls. Keep investigating robot-local scheduler/status, startup reconnect state, CPU/load, and log/upload work if backup status itself remains sluggish after surprise suppression.
- Test 26 remains the broader regression evidence for gallery success, alarm replacement/delete risk, stop/volume live proof, and short-answer STT weakness. Alarm replacement/menu agreement is still the main release risk after the Test 27 cloud-version-tail hardening.
For `1.0.18`, the remaining release work should stay small: finish one or two feature slices, run the live regression pass, and only patch bugs found in that pass before calling the version complete. `1.0.19` should then reopen the broader feature queue.
- passive local context cleanup for gallery/create/settings contexts after stock local skills take ownership
- no-input local completion for constrained prompts, clock value prompts, gallery preview prompts, and settings volume-control prompts
- active local prompt preservation so `shared/yes_no`, clock, gallery, and settings prompts can still consume transcript-bearing short replies even when the stock skill reports a local context
- binary audio ignored for an existing transID until a fresh `LISTEN` has been seen, preventing context-only or post-speech tails from reopening an endless buffered turn
- blank-audio hotphrase turns clear pending listen state and install a short late-audio ignore window
-`cloud version` is a one-shot diagnostic: it speaks the version without opening a follow-up turn, then shields the speech tail from self-listen artifacts such as the Test 27 `Cloudford.` capture
- the spoken cloud-version diagnostic avoids saying `Jibo`, because Test 29 showed the prior `Open Jibo Cloud version...` wording could trigger local hotphrase barge-in during Nimbus TTS
- hosted turn results, fallback matches, and local skill redirects now emit `match.skipSurprises = true` so stock BE does not route settled cloud/local responses into `@be/surprises`
- apostrophes are no longer escaped to `'` in spoken ESML, while `&`, `<`, `>`, and `"` remain escaped
- radio voice launch supports `open the radio` and genre launch such as `play country music`, using local `@be/radio``menu` payloads, `SKILL_REDIRECT`, and silent completion
- news has a first Nimbus-shaped cloud path using `match.cloudSkill = news` and a `news``SKILL_ACTION` with synthetic briefing content
- stop commands such as `stop that` and `never mind` emit stock `global_commands``stop` NLU plus a local `@be/idle` redirect, without generic chat speech
- short clock value follow-up transcripts are accepted under `clock/alarm_set_value` and `clock/timer_set_value` instead of being dropped before parsing
- photo flows route `open photo gallery`, observed `open photogal`, `snap a picture`, and `open photobooth` to the matching gallery/create local skills
- passive gallery/create/settings context does not reopen a stale cloud turn
- active local prompts under gallery/settings context are preserved so short `yes`/`no` answers can finalize the prompt instead of being suppressed as passive context
- no-input fallback for constrained yes/no prompts emits local `LISTEN`/`EOS` instead of relaunching generic Nimbus speech, including `shared/yes_no` after STT failure
- no-input fallback for clock value prompts, `gallery/gallery_preview`, and `settings/volume_control` emits local `LISTEN`/`EOS` instead of generic `I heard you` Nimbus speech
- repeated empty `create/is_it_a_keeper` replies redirect to `@be/idle` after the second miss so the photo/create flow can settle instead of leaving a stale listening state
- local whisper skips buffered audio turns that do not contain `OpusHead`, preventing a known `ffmpeg` failure path from becoming the noisy failure mode
- Word of the Day launch, spoken guesses, structured `CLIENT_NLU` guesses, hint-order guesses, fuzzy hint matching, right-word cleanup, and late audio cleanup are covered in the websocket layer
The Pegasus tree is especially useful for cloud service intent: `packages/hub` documents `/v1/listen`, `/nlu`, and `/asr`; `packages/lasso` documents credential and provider aggregation; `packages/history` and the architecture materials are useful for future memory and proactivity work.
The JiboOS trees are especially useful for local skill ownership and payload shape: `@be/clock`, `@be/gallery`, `@be/create`, `@be/radio`, `@be/nimbus`, `@be/settings`, `@be/surprises*`, `@be/restore`, `@be/who-am-i`, and `@be/idle`.
The original test suites are useful as behavior contracts before more live-device trial and error:
-`..\jibo\sdk\skills\clock\tests\AlarmTimer` documents alarm/timer state expectations. Cancel at the alarm value prompt exits without scheduling; no-alarm query `yes` redirects to the value prompt while `no` exits without touching KB/scheduler; existing-alarm `keep` preserves KB/scheduler while `delete`, `change`, and `cancel` clear it; cross-domain cancel uses the `OtherSet` yes/no branch before deleting the other clock domain.
-`..\jibo\sdk\skills\gallery\tests` documents gallery ownership. Empty gallery `yes` redirects to `@be/create`, empty gallery `no` exits, media-load failure exits, gallery/item views lifecycle out around two minutes, and delete confirmation only deletes on a positive `yes`.
-`..\jibo\sdk\skills\surprises-ota\tests\OTASurprise.test.js` shows OTA/backup surprise priority is robot-local and rate-limited by status plus last-notification timestamps. Backup-in-progress sluggishness should be investigated as local scheduler/status behavior before assuming a cloud backup API issue.
-`..\jibo\sdk\skills\nimbus\tests` and `..\jibo\pegasus\packages\integration-tests-int\src\listen*.test.ts` show the cloud/Nimbus contract: listen transactions emit `SOS`, `EOS`, and `LISTEN`, with optional `SKILL_ACTION`; matched responses preserve `match.skillID` or `match.cloudSkill`; `CLIENT_ASR` and `CLIENT_NLU` should both be first-class test inputs.
-`..\jibo\pegasus\packages\report-skill\tests\subskills\News.test.js` is the best source-backed guide for news expansion: use category preferences, filter unusable or duplicate items, gate adult headlines for children or unidentified speakers, and provide image metadata alongside spoken headlines.
When sources disagree, prefer the newest live stock-OS capture for runtime behavior, then stock robot source for local ownership, then Pegasus for original cloud intent, then Node for known working compatibility behavior.
- Confirm the running robot build reports cloud version `1.0.18` using the shorter `Cloud version ...` wording, without stopping itself on a hotphrase, reopening a late `LISTEN`, or producing a follow-up `Cloudford` / generic chat tail.
- Confirm cloud-version and one generic Nimbus/chat turn include `match.skipSurprises = true` and do not transition into `@be/surprises` / `@be/surprises-ota` after speech completes.
- Regression test alarm flows again after the `jibo test 26` fixes: set with explicit time, set with compact/spoken/comma-separated time, clarify missing time, replace an existing alarm, cancel/delete by voice including `delete the alarm`, cancel out of a value prompt, and verify the menu agrees.
- Regression test timer flows after the Test 25 stale-timer observation: set a 10-second timer, let it fire, reset by gesture only after recording state, and verify a new timer prompt does not see an already-expired timer as still active.
- Regression test photo/gallery flows again after the `jibo test 26` fixes: open gallery, answer the stock `shared/yes_no` prompt with a transcript-bearing `yes`, hand into create, take one photo, keep it, and avoid blue-ring, `I heard you`, or `that's` stale turns after gallery cleanup.
- Regression test the added stop and volume slices after the Test 26 fixes: `stop that`, `never mind`, `turn it up`, `turn it down`, `set volume to six`, `set volume to 6`, and `show volume controls`.
- Recheck backup/update behavior with explicit attention to robot-local `jibo.scheduler.backupStatus`, CPU/load, log/upload activity, and whether the deployed cloud is involved at all.
- update, backup, and restore are not end-to-end proven, and the `jibo test 22` / Test 26 / Test 27 / Test 28 sluggishness appears tied to robot-local backup status/load, startup reconnect state, or previously unsuppressed end-of-skill surprises
- Tests 27 and 28 showed backup/surprise behavior without corresponding `Backup_*` HTTP traffic; Test 28 isolated the unsuppressed `@be/surprises` lifecycle handoff after Nimbus
- alarm replacement yes/no, alarm voice delete/menu agreement, and long blue-ring cleanup still need successful live proof after the Test 26 and Test 27 source fixes
- remaining stop/volume variants still need live stock-OS proof beyond Test 26's `Never mind.` and `Set Volume 2-6.` passes; robot age and command-versus-question personality routing are not implemented yet
