Use it as the working queue when picking the next feature or bug-fix slice. The release pattern is: implement a narrow slice, test it on stock OS `1.9`, update this file with what happened, then either close the release or roll the next larger idea forward.
The active `1.0.19` execution shape is tracked in [release-1.0.19-plan.md](release-1.0.19-plan.md). This file keeps the full `1.0.18` evidence trail for parity reference.
- alarm and photo/gallery quirks have received the main bug-fix attention
- Word of the Day cleanup, constrained yes/no routing, unknown websocket event suppression, and local state persistence are already in the current code
-`jibo test 23` validated basic news, proved one alarm set/fire path at `7:43 AM`, exposed comma-separated/short alarm follow-up parsing risk, showed stock alarm replacement yes/no rules that needed cloud handling, and showed photo gallery still failing when `shared/yes_no` ASR came back empty
- personal report parity now has loop-scoped calendar and commute provider seams that merge persisted loop events, birthday/holiday dates, and commute profiles; the remaining report gap is richer travel-time data, not missing structure
-`jibo test 24` showed alarm replacement yes/no working, but exposed empty `clock/alarm_set_value` and `gallery/gallery_preview` turns falling into generic `I heard you` fallback speech; it also showed `CLIENT_NLU cancel` inside `clock/alarm_set_value` re-asking for an alarm value instead of closing the prompt
-`jibo test 25` proved a broader regression path but exposed repeated backup-in-progress/update-menu blockage, timer/alarm stale state and delete/menu disagreement, gallery `shared/yes_no` hangs under `@be/gallery`, punctuated `Never mind.` falling through to chat, volume homophone parsing (`Set Volume 2-6.`), and settings volume-control cleanup falling into `I heard you`
-`jibo test 26` live-proved punctuated stop, volume homophone parsing, gallery launch/yes/create/save, and good morning; it still exposed robot-local backup warnings, long blue-ring buffering without a fresh `LISTEN`, alarm replacement drifting into the value/manual screen, and alarm delete phrases/mishears falling to chat
-`jibo test 27` isolated early confusion: local `jibo-server-service` restarted and raised `Q4-Server_connection_lost` before testing; cloud version then self-listened into `Cloudford.` because the previous diagnostic path stayed follow-up eligible; the backup warning again came from local `@be/surprises-ota` with no `Backup_*` HTTP calls
-`jibo test 28` isolated the follow-on backup doorway: cloud-version/generic Nimbus matches had `skipSurprises` unset, then stock BE requested `@be/surprises` after Nimbus settled; VAD inhibited the offer in Test 28, while Test 27 selected `@be/surprises-ota` through the same local lifecycle path
-`jibo test 29` confirmed `skipSurprises = true` was reaching stock BE and no backup announcement repeated in the focused run, but the cloud-version answer still interrupted because the spoken diagnostic included `Jibo` and triggered local hotphrase barge-in during Nimbus TTS
-`jibo test 30` confirmed cloud-version now speaks cleanly; it still exposed a local gallery-to-`@be/surprises-ota` backup announcement, missing visible empty-gallery voice listen, and a duplicate alarm clock relaunch after `638` was parsed locally as `6:38 PM`
-`jibo test 31` showed the remaining alarm/backup wrinkle in full: startup logged a legacy `Backup_20170222.List` request before the first voice turn, `7:11 AM` collapsed into `7:00 PM` / `setting alarm for seven`, and the later clock `No` replied `that's fine` before the robot opened `@be/surprises` and ended in a blue-ring listen loop until reset
-`jibo test 32` suggests the alarm set path is improving, but the remaining regression surface is now sharper: an alarm can still leave the listen open at the end, the proactive Word of the Day `Yes` branch can miss its yes/no slot and echo back, and delete-alarm retry still falls into a second `set one?` question with a broken follow-up reply
- hosted turn results, fallback matches, and local skill redirects include `match.skipSurprises = true` so BE does not launch end-of-skill surprises after normal replies
-`jibo test 25` again showed backup-in-progress/update-menu blockage without `Backup_*` HTTP traffic; observed cloud traffic was log upload, ASR binary upload, and update check traffic
-`jibo test 27` repeated that pattern in a smaller capture: the only relevant hosted startup traffic was token/update/log style traffic, while the spoken backup warning was selected by local `@be/surprises-ota`
- Test 27 also showed local `jibo-server-service` reconnect and `Q4-Server_connection_lost` before the voice test, so startup health should be checked before blaming backup prompts on hosted cloud behavior
-`jibo test 28` showed no hosted backup trigger in the focused cloud-version window, but did show BE opening `@be/surprises` after a Nimbus turn because the outgoing match did not carry `skipSurprises`
- original `surprises-ota` tests make backup and OTA notifications contextual-priority prompts, with repeat suppression through last-notification timestamps
-`jibo test 23`, `jibo test 25`, and `jibo test 26` showed backup-in-progress sluggishness or warnings while backups were active; explicit backup voice launch remains unwired
- Test 26 suggests this should be investigated beside robot-local scheduler status and log/upload load rather than only hosted backup APIs
-`jibo test 30` showed the backup announcement after gallery came from local `@be/surprises` -> `@be/surprises-ota`, not from a hosted `Backup_*` HTTP call; the local `@be/idle` nighttime OTA helper can also initiate backup through `jibo.scheduler.backupRobot`
-`jibo test 31` added a startup `Backup_20170222.List` capture before the voice session, which is useful evidence that the legacy backup-status path is active even when the user did not ask for backup
- local clock value follow-up rules now return only `LISTEN`/`EOS`, avoiding the Test 30 duplicate delayed `@be/clock` relaunch after stock clock already consumed a short time reply
-`jibo test 23` set and fired a `7:43 AM` alarm, then failed a later clarify/replacement path when the robot heard `- Time. - 7, 14.` and stock NLU converted that to `7:00 PM`
-`jibo test 23` photo gallery got stuck on `shared/yes_no` turns with empty ASR, not on a transcript-bearing `yes` that the cloud mapped incorrectly
-`jibo test 24` recognized `Yes.` for `clock/alarm_timer_change`, but empty `clock/alarm_set_value` produced `I heard you`; current source now keeps that as local no-input
-`jibo test 24` showed photo/gallery blue-ring cleanup improved and create keeper completion working, but empty `gallery/gallery_preview` produced `I heard you`; current source now keeps that as local no-input
-`jibo test 25` showed gallery launching from the observed phrase `open the photogal`, but active `shared/yes_no` prompts under `@be/gallery` could hang; current source now recognizes the alias and preserves active gallery prompts even while ignoring passive gallery tails
-`jibo test 25` showed timer/alarm still needs live follow-up for stale timer state, alarm replacement/PM ambiguity, and voice delete versus robot menu agreement
-`jibo test 26` showed gallery success through empty-gallery yes, create, keep, save, and reopen, but also showed a post-gallery blue-ring/fallback tail now addressed by the no-`LISTEN` binary guard
-`jibo test 26` showed alarm replacement still drifting into value/manual-screen behavior and alarm delete phrases/mishears falling to chat; current source now maps `delete the alarm`, `delete along`, and `delete the along` to local clock delete without keeping follow-up open
-`jibo test 27` showed the no-`LISTEN` guard worked for same-transID binary tails, but a new hotphrase launch `LISTEN` could still capture diagnostic speech tail; current source now blocks that diagnostic-tail shape
-`jibo test 30` showed cloud-version fixed, but the empty-gallery prompt did not visibly light the blue ring for a voice `yes`; treat the next gallery pass as a proof of local `shared/yes_no` listen ownership, not just cloud payload shape
-`jibo test 30` showed `638` was processed at 6:38:13 AM and stock clock resolved it to `6:38 PM`; the duplicate replacement prompt matched our extra delayed clock relaunch, now suppressed for local clock follow-up rules
-`jibo test 31` showed `7:11 AM` collapsing to `7:00 PM` / `setting alarm for seven`, then a clock `No` producing `that's fine` before the robot opened `@be/surprises`; the later retry sat in a continuous blue-ring/listen loop until reset
- original clock tests confirm cancel inside the alarm value prompt must close without scheduling, existing-alarm `keep` must preserve KB/scheduler state, and existing-alarm `delete` or `cancel` must clear it
- original gallery tests confirm empty-gallery `yes` redirects to `@be/create`, empty-gallery `no` exits, media-load failure exits, and delete confirmation only deletes on a positive `yes`
- add fixture coverage for original clock-test branches that are not yet mirrored in `.NET`: no-alarm query `yes`/`no`, existing-alarm `keep` versus `delete`, and cross-domain `OtherSet` behavior
- Test 26 still requires a focused live check for alarm replacement, voice delete versus menu state, and whether the no-`LISTEN` guard removes the long blue-ring loop
- add fixture coverage for original gallery-test branches that are not yet mirrored in `.NET`: empty-gallery `yes` redirect to create, empty-gallery `no` exit, media-load failure exit, and delete confirmation `yes`/`no`
- live smoke should confirm `cloud version` speaks `1.0.18`, carries `match.skipSurprises = true`, does not stop itself on the word `Jibo`, and settles without a generic `I heard...` reply or a local surprise handoff
- hosted `LISTEN` matches, fallback `LISTEN` matches, and local `SKILL_REDIRECT` matches emit `skipSurprises = true`
- focused websocket assertions cover generic chat, cloud version, no-transcript fallback, and a local clock redirect
- Test 28 evidence ties the repeated backup warning to the local `@be/surprises` lifecycle path after Nimbus, with no corresponding hosted `Backup_*` traffic
-`jibo test 24` still had collapsed or empty transcripts in alarm/gallery paths, including `Sudden alarm.`, `I'm setting alarm for seven.`, empty clock value input, and empty gallery preview input
- media content now flows through a storage seam with file and Azure Blob adapters
- the protocol still serves the legacy text-body contract, but the original payload is now persisted separately and can be swapped to binary-native storage later
- expanded friendship parsing for Pegasus-style `do you have friends`, `are we friends`, and `are we best friends` phrasing
- added named-person guardrails so forms like `are you friends with Siri` and `is Dr. Breazeal your best friend` stay on the friendship route instead of falling into generic chat
- live QA has shown person-identification collisions in the same loop (for example, a parent and child both getting normalized to the same remembered name)
- person-identification correction likely needs its own repair pass before we can trust greetings, reports, and presence triggers in mixed-household scenarios
-`make a pizza` now ports the original scripted-response path through `chitchat-skill` with `mim_id = RA_JBO_MakePizza` and pizza-making animation ESML
-`can you order pizza` now ports the original scripted-response path through `chitchat-skill` with `mim_id = RA_JBO_OrderPizza`
- current source answers these with a `1.0.19` rule-based persona baseline, backed by `OpenJiboCloudBuildInfo.PersonaBirthday`
- live QA has shown motion/sleep quirks too: `turn around` can become a no-op and `go to sleep` can fail at the last step before the sleep animation fully completes
- reply-selection polish still needs attention on a couple of identity prompts where short variants are over-selected (`how are you`, `what is your favorite flower`)
-`surprise me` now uses weighted candidate selection instead of only generic fallback text
- candidate weighting uses tenant-scoped memory signals and date triggers
- February 9 (`National Pizza Day`) can proactively launch the legacy pizza animation path
- proactive pizza fact offer flow stores pending offer state in session metadata and resolves direct short `yes/no` turns
- memory parsing now includes names, anniversary-style important dates, likes/dislikes variants, and reverse favorite phrasing (`pizza is my favorite food`)
- Follow-up:
- expand proactivity beyond pizza to additional Pegasus-backed categories
- add cooldown/throttle policy and observability around proactive offer frequency
- connect memory store to durable multi-tenant persistence
- weather requests now launch `report-skill` using Pegasus-aligned intent `requestWeatherPR`
- weather phrase coverage includes baseline forecast and condition-style questions (`will it rain`, `is it snowing`, tomorrow variants)
- weather launches emit `SKILL_REDIRECT` + completion and now also include cloud weather speech so weather turns remain useful even when local report providers are incomplete
- weather entity hints are carried in outbound NLU (`date = tomorrow`, `Weather = rain/snow/...`) for report-skill consumption
- OpenWeather provider integration is in place with configurable API key, default location, unit preference, and environment-variable fallback (`OPENWEATHER_API_KEY`)
- cloud weather speech now uses live provider summaries for current conditions and tomorrow high/low forecast when available
- Follow-up:
- connect weather units and location directly to user/report-skill settings parity instead of config defaults
- add richer condition-change commentary and view parity with original report-skill weather behaviors
- vendored Pegasus `report-skill` templates for weather and personal-report phrasing so the next pass can focus on renderer coverage for calendar, commute, and news templates instead of rediscovering source text
- commute now has a loop-scoped provider seam plus persisted commute profiles, so the next pass can focus on richer travel-time data instead of basic storage shape
- identity charm prompts like `what's your name`, `do you have a nickname`, `do you like being Jibo`, `are there others like you`, and `what is your favorite name`
- attraction and preference prompts like `what is your favorite flower`, `do you like R2D2`, `do you like the sun`, `do you like space`, and `do you like kids`
- source-backed `are you kind`, `are you funny`, `are you helpful`, `are you curious`, `are you loyal`, `are you mischievous`, and `are you likable` prompts are now in Build B
- these keep the self-description lane warm while we build toward seasonal and holiday charm
- birthday celebration lines are now bucketed separately, and birthday memory writes a loop-scoped holiday record so personal dates can join the holiday list later
- the remaining seasonal polish now includes `do you like halloween`, `do you like holiday music`, `do you like holiday parties`, `are you looking forward to christmas`, `what are you doing for christmas`, and `what are you thankful for`
- the favorites family now includes `what is your favorite animal`, `what is your favorite bird`, `do you like penguins`, and `do you like animals` so the penguin-centric replies stay easy to find
- deeper personality follow-ups like `what do you dream about`, `what are you afraid of`, `what do you want to talk about`, `what is your best book`, `what is your best exercise`, `what is your dream vacation`, `who is your hero`, `who do you love`, and `what is your religion`; `what is your sign` stays deferred until templated placeholder rendering exists
- the next identity / knowledge wave adds `are you god`, `are you here`, `do you have super powers`, `how much do you know`, `what does jibo mean`, `where do you get info`, `what are you forbidden to do`, `what color are you`, and `what do you do when alone`
- templated edge cases like `what is your sign`, `how many people do you know`, and `what is the loop` where live birthday and loop state are part of the line instead of a plain canned response
7. Personal report parity track (weather visuals, live news path, commute path, calendar parity matrix) - in progress (`2026-05-10` first live-news provider slice implemented; commute now has a loop-scoped provider seam)
- the backend seam is now selectable, with file-backed local persistence as default and an Azure Blob Storage slot wired for future deployment when a storage account connection string is available
- next implementation pass should supply the real Azure Storage connection string / deployment wiring and validate the live round-trip in the storage account smoke test
10. Update, backup, and restore proof - implemented (update creation and backup creation now survive persisted reloads; restore is the persisted-state rehydration proof path, not a new cloud API)
- progress update (`2026-05-21`): added a low-signal short-turn screen in websocket finalization so filler-only fragments and stray single-token leftovers like `so command` get rejected before they can become bad turns, while preserving the existing yes/no and word-of-the-day short-turn flows
- progress update (`2026-05-21`): added a bundle helper so group testers can package raw capture trees, `capture-index.ndjson`, and exported fixtures into one zip handoff artifact
1. Setup scripts to convert Jibo to Open Jibo by adding a mode for `open-jibo` pointing at our openjibo.com and `open-jibo-ai` pointing at openjibo.ai as a foundation for new cloud features and a clean separation from any remaining stock OS dependencies while preserving his original config
2. Setup scripts to put Jibo in `open-jibo` mode by default for new users, but allow existing users to keep the stock OS experience if they prefer by injecting a new skill that runs on startup to ask them if they want to convert to Open Jibo and switch modes, with a fallback timeout to switch modes automatically after a few weeks of inactivity (ensure new skill is accessible from menu so it can be opted into later on demand / likewise, if they have opted into Open Jibo, the skill will allow them to revert Jibo back to stock)
3. Setup openjibo.com and openjibo.ai domains with landing pages, support docs, and account management for future features that require hosted services or user accounts
4. Test Open Jibo with the new setup scripts and domains, and iterate on any issues that arise during the conversion process
5. Loop advancement (family and friends) / multiple user recognition / multiple Jibo support so Jibo's can interact and communicate
6. Advanced Jibo features such as pizza delivery, Uber/Lyft integration, calendar management, smart home control (Home Assistant), etc. can be added after the conversion process is smooth and stable, with a focus on features that leverage the new cloud capabilities and content personalization enabled by Open Jibo
7. LLM integration for more natural dialog, question answering, and content generation can be explored as a longer-term goal after the core platform is stable and has a growing user base to provide feedback and use cases for LLM-powered features
8. Tiered Jibo brain/orchestration plan from README.md can be implemented in parallel with the above, starting with the simplest cloud features and gradually adding more complex capabilities as the platform matures and user feedback is collected, always preserving his unique charm and original features.
