ISEKAIBONDS
Google Play

Illustrated guide · V0.21.26 reference

The complete guide, screen by screen.

Install Isekai Bonds from Google Play. This illustrated guide and the English PDF document V0.21.26; some screens and limits may differ from the current release. Personal VRM and image imports require Pro.

Get it on Google Play
71 sections69 illustrated steps Validated source package
Official English PDF143-page accessible A4 edition

01 · First launch

Open the portal

Steps

  1. Read Choose your companions, Start your story, and Write or dictate.
  2. Use Next, or Skip, then Open Chat.
  3. Locate Chat, Quick guide, My characters, AI models, Avatar Studio, and Settings.
Isekai Bonds home with Chat, Quick guide, and My characters
Official Home screenshot.

00 · Starting

Wait for local preparation

Steps

  1. Launch Isekai Bonds and let the bar reach 100%; do not close the application during the initialization of local stores.
  2. At the very first launch, the preparation screen is voluntarily in English before the registered language is known.
  3. If the home does not open, restart once and then attach the displayed message to your report without including any private data.
Authentic start screen while preparing local services.
Authentic start screen while preparing local services.

01B · Quick Guide

Step 2 - Launch Your Story

Steps

  1. After the first card, tap Next.
  2. Read the available accesses to start a conversation and select a companion.
  3. Return with Previous if you want to read Step 1.
Second authentic step of the quick guide.
Second authentic step of the quick guide.

01C · Quick Guide

Step 3 - Write or dictate

Steps

  1. Touch Next from step 2.
  2. Find the text input and Android dictation described in the map.
  3. Finish with Open Chat or use Pass to directly reach conversations.
Third authentic stage of the quick guide.
Third authentic stage of the quick guide.

02 · Characters

Choose or create a companion

Steps

  1. Open My characters and first try a built-in character with Start a chat.
  2. Tap New character and enter at least a name, role, and personality.
  3. Optionally add appearance, style, opening messages, a preferred local model, and voice.
  4. Save and confirm that the character appears in the library.
Library of built-in and custom characters
Authentic character library.

04 · Profile and avatar

Edit a character without losing chats

Steps

  1. Open My characters, choose Edit, then review identifier, name, role, language, personality, backstory, and opening messages.
  2. Choose Static image or Animated VRM avatar. For VRM, select a built-in avatar or open the private library.
  3. Set the preferred local model, generation profile, and voice, then Save changes.
  4. Create a new conversation: the saved avatar and portrait appear from the first chat screen.
  5. Duplicate creates an independent profile. Delete asks for confirmation and also removes that character’s local conversations.
Character profile with avatar and model choices
The profile links identity, avatar, model, and voice.

03A · Appearance

Separate images by emotion

Steps

  1. Open Characters → Edit → Images and emotions.
  2. Choose An Emotional Image and select the emotion to inform.
  3. Import only an image with your rights, adjust the framing, test and save.
Direct screen Images and emotions in emotion mode.
Direct screen Images and emotions in emotion mode.

03B · Appearance

A unique image for all emotions

Steps

  1. In Images and Emotions, open the mode menu.
  2. Select Unique Image, import the allowed image and check its framing.
  3. Record; the same portrait will be used at the beginning of the new chats and for all emotions.
Direct screen Images and emotions in single image mode.
Direct screen Images and emotions in single image mode.

05 · Conversations and history

Search, resume, rename, or delete

Steps

  1. Open Chat: each card shows character, title, excerpt, type, and date.
  2. Enter a term in Search conversations to filter, then Open to resume the exact local thread.
  3. Rename changes only the card title; messages and memory stay intact.
  4. Inside a chat, History opens every message. Close history to return to the same conversation.
  5. From the list, Delete opens a confirmation. Cancel changes nothing; Delete permanently removes the conversation and its local history.
Permanent conversation deletion confirmation
Deletion requires explicit confirmation.

03 · Solo

Solo chat and solo roleplay

Steps

  1. Open Chat, then New conversation.
  2. Select a character, name, and Direct chat for a face-to-face conversation.
  3. Select Narrative roleplay for a narrated scene and optionally apply a reusable scene.
  4. Send a message. History restores exchanges; three suggestions offer possible replies.
New conversation dialog with character and mode
All four conversation types use this dialog.

09 · Message actions

Act on one exact point in history

Steps

  1. In History, tap Actions on the relevant message.
  2. Copy text places only that message on the clipboard.
  3. Pin as memory and Mark as important event enrich user-controlled memory.
  4. Edit then continue rewrites that point and rebuilds the continuation; Create a branch here preserves the original and opens a derived conversation.
  5. Illustrate this moment prepares an image from that message. Report this content appears only on AI replies.
Message Actions menu in History
Actions apply only to the selected message.

04C · Studio avatar

Test the avatar prototype

Steps

  1. Open Studio → Test theavatar from the library.
  2. Wait for the VRM load, then check the settings, rotation, animation, expressions and labial synchronization.
  3. If the rendering remains empty, go back to the library, reactivate theavatar and check its provenance/licence.
Direct VRM prototype test screen.
Direct VRM prototype test screen.

06A · New chat

Choose the mode actually available

Steps

  1. Touch Chats → New conversation.
  2. Open the Type menu: Direct Discussion, Narrative Roleplay, Group Discussion or Group Roleplay.
  3. Select the character(s); the direct group and the Roleplay group are subject to the Free/Pro limits displayed.
  4. Validate and check that the Open Screen Type language matches your choice.
Direct menu of the four active conversation modes.
Direct menu of the four active conversation modes.

06B · Conversations

Rename without losing history

Steps

  1. In Chats, find a conversation card and tap Rename.
  2. Replace the selected title with a non-sensitive name and then tap Rename.
  3. Reopen the conversation and check that its messages, memory, journal and world are unchanged.
Direct French window renaming a conversation.
Direct French window renaming a conversation.

07A · Background

Check and retrieve multiple messages

Steps

  1. In a conversation, touch History.
  2. Scroll several tricks and find author, text, possible date and associated illustration.
  3. Tap Resume/Open Message when the action is proposed; the conversation returns to this point without exposing any other history.
Direct history opened with several messages.
Direct history opened with several messages.

07B · Message actions

Compare three variants

Steps

  1. After a complete answer, tap 3 answers.
  2. Wait for the three proposals and read them before choosing; this action can consume an additional cloud or local call.
  3. Touch a variant to place it in the draft, edit it if necessary and then send it.
Direct window containing three variants of response.
Direct window containing three variants of response.

07C · Message actions

Create a safe branch

Steps

  1. Open History, choose a compatible message and tap Connect.
  2. Read the confirmation: the branch copies the messages to the selected point and can copy memory, journal, world and private images.
  3. Confirm, give a title to the branch and check that a new card appears in Chats.
Direct window creating a branch.
Direct window creating a branch.

07D · Message actions

Check the branch created

Steps

  1. Return to Chats after confirmation.
  2. Open the new card marked Branch and review the last copied message.
  3. Compare Memory, Journal and World with the original before continuing a different timeline.
Authentic trendy conversation after creation.
Authentic trendy conversation after creation.

04 · Group and game

Bring several characters together

Steps

  1. Group chat accepts two or three same-language characters. Speaker can stay Dynamic - based on messages or force one companion.
  2. Pass turn changes the respondent without a fake player message or World Master.
  3. Group roleplay adds Direction: dynamic by scene, a forced character, or World Master - scene and NPCs.
  4. Continue advances the scene without creating a fake player message.
Group roleplay with Direction and World Master
Authentic group-roleplay capture.

07 · Streaming

Watch the response arrive

Steps

  1. Select Local GGUF model and load its weights.
  2. Open a conversation and send a short message.
  3. Confirm that text appears progressively and the status names the local model.
  4. Tap the ■ square: generation stops and received text remains visible.
Local response appearing progressively with a stop button
Authentic GGUF streaming in chat.

12 · Context, memory, and continuity

Control what guides the next reply

Steps

  1. Open More then Memory to review pinned memories, important events, and scene state.
  2. Context preview details character, mode, provider, recent conversation, older memories, World, journal, and budget prepared for the AI.
  3. Refresh recomputes the preview with the current draft without sending a request.
  4. Relationship adjusts trust, affection, and tension; Journal tracks quests, decisions, encounters, and items; World stores active canonical facts.
  5. Save each panel after editing. Changes affect subsequent replies.
Transparent preview of the next-reply context
The panel shows what will actually be prepared for the AI.

05 · Cloud LLM

Configure your provider

Steps

  1. Open the AI model section in Settings.
  2. Choose OpenAI, Google Gemini, Anthropic Claude, or an OpenAI/OpenRouter-compatible service from the shown presets.
  3. Enter your own key, model, and a base URL only when required.
  4. Enable secure storage if desired, then Save and Test before sending a short message.
Cloud model provider selection
Capture of the actual provider list.

12A · Cloud images

Configure the provider without exposing the key

Steps

  1. Open Settings → Create cloud images.
  2. Choose the proposed provider and model; For Cloudflare, provide the Worker's HTTPS URL, for others the service requests its dedicated key.
  3. Enable the Android Keystore, save and then tap Test Connection.
  4. Never capture the key field; The "available protected key" status is sufficient in a report.
Direct configuration of the image generation, hidden key.
Direct configuration of the image generation, hidden key.

12B · Cloud images

Trigger from Scene

Steps

  1. In an active conversation, touch Scene.
  2. Describe the setting or timing; check the quota and transmission warning.
  3. Touch Generate image once and then leave the window open.
Direct French window before generation of an image.
Direct French window before generation of an image.

12C · Cloud images

Wait for progress

Steps

  1. During generation, observe the progress message and the Generate button disabled.
  2. Do not send a new message and do not force closure during the query.
  3. Wait up to two minutes depending on the provider and network.
Direct image generation status in progress.
Direct image generation status in progress.

12D · Cloud images

Review, save, or delete the result

Steps

  1. Check that the illustration corresponds to the moment without unexpected private content.
  2. Tap Fullscreen to inspect, then Save Image for a selected copy; The original remains in private storage.
  3. The thumbnail attaches to the message and becomes visible in History and, in Pro, Gallery of Adventures.
  4. To delete, use the image action or delete the message/conversation after confirmation.
Genuine result of a generation of images in conversation.
Genuine result of a generation of images in conversation.

12E · Cloud images

Handle errors without losing the chat

Steps

  1. Read the exact message: network, HTTP, quota, model, key or response format.
  2. Close the window, correct Settings → Create cloud images and test the connection.
  3. Open Scene; the draft and historical remain local, then restart once.
Direct network error during image generation.
Direct network error during image generation.

13A · Local models

Import or search for a GGUF

Steps

  1. Open Models IA; The empty state proposes Import a GGUF and Hugging Face.
  2. For a local file, choose an authorized .gguf: extension, size and signature are validated before private copying.
  3. For Hugging Face, look for a GGUF public repository, compare size, quantification, estimated RAM and license, and then download the selected file.
Direct GGUF library in empty state.
Direct GGUF library in empty state.

13B · Local models

Load, use and unload

Steps

  1. Select a card and then tap Load and use or Load in Memory.
  2. On Android, follow Loading → ready; OpenCL is tried if compatible, otherwise the message announces the automatic CPU fold.
  3. Choose the default or character model, then run a chat and wait for the initial context loading.
  4. Return here and tap Unload before a heavy local TTS or to release RAM; Delete erases the file after confirmation.
Direct GGUF library with models, selection and loading commands.
Direct GGUF library with models, selection and loading commands.

13C · Local models

Understand why a GGUF import was rejected

Steps

  1. Tap Import a GGUF and choose a file actually carrying the .gguf extension; the native system selector normally filters the other files.
  2. After selection, wait for "copying and validation of GGUF...": the application checks extension, maximum size of 32 GB and GGUF header before any private copy.
  3. If the status displays "Select a .gguf file", return to the selector and choose the correct file; for "Unable GGUF header", redownload the model from its official source.
  4. For a file too large or incomplete, free space, check its published checksum and prefer a Q4 quantification from 1B to 3B adapted to the phone.
Direct French screen after refusal of a non-GGUF file by the actual validator.
Direct French screen after refusal of a non-GGUF file by the actual validator.

06 · Local GGUF

Download, import, and load a model

Steps

  1. From Home, open AI models and the local/GGUF library.
  2. Import a legally obtained .gguf file or use public Hugging Face search without a token.
  3. Start with a small Q4 suited to device memory and wait for the complete download.
  4. On its card, use Set as default, then Load into memory or Load and use.
GGUF model library with loading actions
An installed GGUF can work offline.

08 · Voices

Android, Piper, Supertonic, and cloud

Steps

  1. Enable voice in a character profile, then open Voice library.
  2. Select a language, download a Piper or Supertonic voice, and play its preview.
  3. For Supertonic, read and accept the OpenRAIL-M license before downloading.
  4. Assign the voice, return to the profile, and use Test voice.
Voice library with Piper and Supertonic sections
Authentic local voice library.

14A · TTS

Activate voice on character profile

Steps

  1. Open Characters → Edit and descend to Voice of Character.
  2. Activate voice, choose Android, Piper, Supertonic 3 or a cloud provider actually configured.
  3. Select voice and language, set speed, height if the engine supports it, volume, auto playback, narration and interruption.
  4. Touch Test Voice; correct the selection if no sound starts.
Direct TTS commands of the character profile.
Direct TTS commands of the character profile.

14B · Local TTS

Read license, language and memory before Piper

Steps

  1. Touch Voice Library.
  2. Find SIWIS - English, en-FR, recommended size, RAM and CC-BY-4.0 license.
  3. Touch Download only after checking the license and free space.
Direct Piper library with language, size and license.
Direct Piper library with language, size and license.

14C · Local TTS

Follow or cancel Piper download

Steps

  1. During download, check the percentage.
  2. Use Cancel download if network becomes unstable.
  3. After 100%, wait for the check sum validation and installation.
Download 98% Piper Direct.
Download 98% Piper Direct.

14D · Local TTS

Assign and preview Piper

Steps

  1. In Local Voice installed, tap Listen for an overview.
  2. Tap Use to assign to the current character.
  3. Go back to the file, check Piper/SIWIS and then test the voice.
  4. Use Delete only if no character has yet to use this voice.
Piper installed with Listen, Use and Delete.
Piper installed with Listen, Use and Delete.

14E · Local TTS

Install Supertonic 3 and accept OpenRAIL-M

Steps

  1. In Voice Library, develop Supertonic 3.
  2. Choose the language; The shared multilingual heart is downloaded once.
  3. Read the OpenRAIL-M package leaflet and explicitly accept its restrictions before downloading.
  4. After installation, assign a Supertonic voice to the character and test it.
Supertonic Packs Direct Library 3.
Supertonic Packs Direct Library 3.

14F · Local TTS

Explicitly validate the Supertonic license

Steps

  1. In Voice Library, tap Download on a Supertonic 3 voice corresponding to the character's language.
  2. In Voice Model License, read the OpenRAIL-M conditions and tap Read the full license to view the text distributed with the template.
  3. Touch J Cancel close the window without downloading.
  4. After acceptance, follow the shared heart and then the voice pack to the installed state, use Listen, then assign the voice from the character file.
Direct French OpenRAIL-M license window before downloading Supertonic.
Direct French OpenRAIL-M license window before downloading Supertonic.

14G · TTS

Play, stop, and combine with streaming

Steps

  1. After a complete answer, tap Replay.
  2. During synthesis/reading, observe the status and use Stop Voice to interrupt.
  3. With Autoplay, the TTS starts after exploitable streaming fragments; a new message may interrupt the voice depending on the setting.
  4. With a local GGUF, the application serializes heavy engines; wait until the TTS is released before a competing local load.
Direct TTS reading with action Stop voice.
Direct TTS reading with action Stop voice.

17 · Voice playback and streaming

Play, stop, and continue without blocking chat

Steps

  1. Enable voice in the profile and choose Automatically play replies if every reply should be spoken.
  2. Text may finish streaming before speech synthesis completes; Preparing indicates background work.
  3. Replay speaks the latest available reply. Stop voice immediately interrupts playback and clears that queue item.
  4. With Interrupt voice on a new message, the next send cuts current playback.
  5. In groups, voice follows the active speaker; World Master may use a separate voice in roleplay.
Group roleplay during voice replay
Stop voice remains available during playback.

09 · Avatar and VRM

Test Studio and an authorized file

Steps

  1. In Avatar Studio, try a built-in avatar, emotions, animations, and lip sync.
  2. VRM import is Pro. Open the library, tap Import VRM, and select an authorized file.
  3. Provide the requested author, source, and license; a file found online is not automatically free.
  4. Select the VRM, tap Activate, then Test avatar before assigning it to a character.
Avatar Studio displaying the built-in Mira avatar
Only a built-in avatar is shown; no third-party VRM is redistributed.

15A · VRM

Import with license and provenance

Steps

  1. Open Studio → Library of d
  2. Choose an authorized .vrm file; The application reads the integrated metadata (title, author and usage rules) instead of requesting a separate form.
  3. After validation, the application copies the VRM to its private storage, creates a thumbnail and displays the available metadata.
  4. If the author or license is empty, keep proof of provenance outside the app and do not use the file until your rights are established; in case of refusal, check format, size, commercial rights and redistribution.
Direct VRM import result with documented test avatar.
Direct VRM import result with documented test avatar.

15B · VRM

Enable, rename and associate

Steps

  1. Select the imported l的avatar in the library.
  2. Touch Enable, then Rename if necessary.
  3. Open Associations to link emotions, lips and movements/BVH to the character.
  4. Check the author and license displayed before any capture use or export.
Direct VRM library with activation commands, testing and associations.
Direct VRM library with activation commands, testing and associations.

15C · VRM

Test rendering, emotions, and lip sync

Steps

  1. Touch Test the avatar.
  2. Try several emotions and movements allowed; Check face, eyes, bones and framing.
  3. Run a vocal test and observe the labial synchronization.
  4. Return to correct Associations or Animations and emotions if an expression is missing.
Direct VRM rendering test.
Direct VRM rendering test.

15D · VRM

Delete with confirmation

Steps

  1. In the library, select an unintegrated avatar.
  2. Tap Delete and read the confirmation on file, thumbnail and associations.
  3. Confirm and check that the associated characters return to an available avatar.
Direct confirmation of VRM deletion.
Direct confirmation of VRM deletion.

20 · VRM library

Import only an authorized avatar

Steps

  1. Open Avatar library then Import VRM file. The picker must receive a valid .vrm file.
  2. Enter or verify name, author, license, and provenance. Do not import a model you are not allowed to use.
  3. Select the avatar and Activate. Rename changes its local label; Test avatar opens the real prototype.
  4. Associations opens expressions and motions. Return to the character profile, select the avatar, and Save.
  5. Delete asks for confirmation. First change any character that uses the avatar.
Private library before VRM import
No third-party avatar is included in the guide.

21 · Emotions, lips, and BVH

Test before saving

Steps

  1. In Associations, choose avatar or character target and one of 28 emotions.
  2. Set VRM facial expression, intensity, duration, and playback mode. System emotions remain available without customization.
  3. Import BVH starts a wizard: choose file, choose an emotion, verify compatibility, test, then save.
  4. Manage BVH lets you enable, loop, choose primary, add to another emotion, move, or delete.
  5. In Studio, run Test all 28, Pause, Next, Stop, and Neutral. Talk with this emotion tests voice, lips, expression, and motion together.
Twenty-eight emotion associations with VRM preview
Settings are separate for each avatar or character.

15E · Associations

Match each emotion to the character

Steps

  1. From Library d的avatars, tap Associations, then choose the Avatar or Character target.
  2. Select one of the 28 emotions and choose “Hériter” (the app’s French label for Inherit) to maintain general behavior, or Customize for this target only.
  3. Match the VRM expression, its intensity, duration and mode of playback; add a BVH only after compatibility validation.
  4. Tap Test before recording, then Save; Delete customization returns to the legacy without erasing shared resources.
Direct screen Associations with distinct target, emotion, face and movement.
Direct screen Associations with distinct target, emotion, face and movement.

15F · Studio emotions

Test 28 emotions and the Talk animation

Steps

  1. Open Studio avatar, choose an authorized avatar and launch Test 28.
  2. Use Pause, Previous, Next, Stop and Neutral to control every face and movement change.
  3. Filter custom or fallback emotions to quickly identify a missing association.
  4. Choose character and emotion, touch Talk with this emotion, control voice and lips, then Stop Talk to interrupt immediately.
Direct studio during the emotion test, distinct from the Associations screen.
Direct studio during the emotion test, distinct from the Associations screen.

10 · Scene Director

Frame roleplay with your local model

Steps

  1. Load a GGUF, make it active, then open Scene Director from Chat.
  2. Copy a built-in example or create a personal scene; Free allows one.
  3. Fill language, name, setting, era, tone, goal, stakes, cast, and introduction.
  4. Save and apply it to Narrative roleplay or Group roleplay.
  5. With a loaded GGUF, speakers and World Master answer sequentially; keep replies short on modest hardware.
Scene Director and built-in examples
Authentic Scene Director.

16A · Stage director

Create a personal scene

Steps

  1. From Chats, open Stage Director and then New Scene.
  2. Inquire title, hanger, frame, atmosphere, conflict, instructions from the Master of the World and visual instructions.
  3. Choose the characters and model, then save.
Direct form for creating a scene.
Direct form for creating a scene.

16B · Stage director

Edit and review the saved scene

Steps

  1. Open the created card, tap Edit and change a field.
  2. Save and return to the list.
  3. Open the scene and check that the fields and meters match.
Direct personal scene after recording.
Direct personal scene after recording.

16C · Stage director

Apply the scene to a conversation

Steps

  1. Touch Use in a conversation.
  2. Choose mode, distribution, characters and model available.
  3. Create the conversation and check that the original context and the Master of the World reflect the scene.
  4. The local model must be installed and ready; Otherwise select a configured cloud provider.
Direct configuration of a conversation from a scene.
Direct configuration of a conversation from a scene.

11 · Settings

Adjust the experience and protect data

Steps

  1. Check language, text size, enhanced contrast, and reduced motion.
  2. Review default solo mode, proactive messages, LLM provider, and generation profile.
  3. Choose local or cloud speech, performance metrics, and scene background.
  4. Export and restore an encrypted .isekai private backup using test data only.
Character profile with personality and voice settings
Character preferences complement global Settings.

24 · Languages, privacy, and backup

Keep control of local data

Steps

  1. Under Application, choose French, English, German, Portuguese (Brazil), or Spanish (Latin America). The change is immediate.
  2. Set text size, high contrast, and reduced motion. No other visual theme is available.
  3. Private backup exports characters, avatars, conversations, Worlds, scenes, and images to an encrypted file.
  4. Enter the same password twice, at least eight characters. It is never stored and the backup cannot be opened without it.
  5. Import requires the file and password. API keys, settings, GGUF models, downloaded voices, and caches are intentionally excluded.
  6. Reset preferences restores settings but does not delete adventures.
Encrypted private backup settings
The backup explicitly excludes keys and downloaded models.

18A · Private data

Export an encrypted backup

Steps

  1. Open Settings → Private backup.
  2. Choose Export, set a highly unused password and keep it out of the report.
  3. Save the archive in a controlled location and check that it exists.
Direct private backup export panel.
Direct private backup export panel.

18B · Private data

Import without unexpected overwrites

Steps

  1. Before import, export the current state.
  2. Choose Import, select the archive and enter its password locally.
  3. Read the summary/confirmation and confirm only if the scope is correct.
  4. Launch the main screens and check characters, conversations, scenes and media.
Direct security check during a backup import.
Direct security check during a backup import.

12 · Reports and feedback

Send only what is necessary

Steps

  1. Long press the exact AI reply.
  2. Open Message actions, then Report this content.
  3. Select a category, add an optional comment, review, and send.
  4. For a general bug, use the private testing-listing channel or support@isekaibonds.com with app version, Android, device, steps, expected, observed, and frequency.
  5. After Send, wait for confirmation of account or network error, then return to the message: the conversation should not be deleted or replaced.
AI-message report dialog
The rest of the conversation remains on-device.

Pro · Google Play

How do I manage a Pro purchase?

Steps

  1. Open Settings and the Isekai Bonds Pro section.
  2. Review the benefits and regional Google Play price before choosing a one-time purchase.
  3. For an existing purchase, use Restore purchases with the same Google Play account. Contact support if the entitlement is missing.
Settings Pro section with an empty reviewer field
This V0.21.26 screenshot contains development controls. It is an archive reference, not proof of a completed purchase or restoration. Public users do not need reviewer credentials.

Pro · Google Play

How do I manage a Pro purchase?

Steps

  1. Open Settings and the Isekai Bonds Pro section.
  2. Review the benefits and regional Google Play price before choosing a one-time purchase.
  3. For an existing purchase, use Restore purchases with the same Google Play account. Contact support if the entitlement is missing.
Pro state direct test in Settings.
This V0.21.26 screenshot contains development controls. It is an archive reference, not proof of a completed purchase or restoration. Public users do not need reviewer credentials.

21A · Dictated Android

Allow microphone only if necessary

Steps

  1. In the Android chat, tap the microphone icon.
  2. Accept RECORD AUDIO if you wish to dictate; otherwise refuse and continue on the keyboard.
  3. If the native service is not available, the app opens the dictation of the Android keyboard.
Direct request for Android microphone permission.
Direct request for Android microphone permission.

21B · Dictated Android

Check recognized text before sending

Steps

  1. During listening, the icon becomes Stop and partial results fill the draft.
  2. Tap Stop or wait for the final result.
  3. Read and correct the recognized text before pressing Send.
  4. In case of silence, denied permission or missing service, use the keyboard and only attach the error message to the report.
Direct state of Android dictation listening.
Direct state of Android dictation listening.

22 · Reading screenshots

Understand the occasional English labels

Steps

  1. The text of the French tutorial describes the current V0.21.26.
  2. Some authentic screenshots come from earlier validation states where French and English were mixed; their legend reports it.
  3. Rely on the section IDs, command position and expected result, then report any non-localized wording in the tested version.
Example of an authentic screen; the legend documents the visible language.
Example of an authentic screen; the legend documents the visible language.

27 · Legacy game mode

Understand what is actually available

Steps

  1. The repository contains legacy New adventure and Portal/roleplay scenes.
  2. In V0.21.26, the isekai feature flag is false: no user-facing button opens those two scenes.
  3. The available game-like path uses New conversation, Group chat or Group roleplay, with Direction and World Master.
  4. Do not edit settings files or bypass feature flags during Google Play testing.

28 · Troubleshooting

Diagnose without erasing data

Steps

  1. Cloud model error: check provider, model, base URL, network, and key, then Test connection. Never copy the key into a report.
  2. Rejected GGUF: verify .gguf extension, complete file, supported quantization, and free space. Try a small 1B-3B Q4.
  3. Load stuck or low memory: unload, choose CPU, reduce context/profile, and retry. GPU remains prioritized for the avatar.
  4. Missing voice: open Library, accept license, finish download, and retest the profile. Supertonic shared core downloads only once.
  5. Rejected avatar/BVH: verify file, license, metadata, and skeleton compatibility. Do not merely rename a wrong extension.
  6. Stuck streaming or TTS: use Stop, wait for Ready, then send a short message.
  7. Before resetting anything, export a private backup and note the displayed version.
GGUF library and load controls
Start diagnosis from the state shown by the app.

29 · Complete tester checklist

Finish a reproducible cycle

Steps

  1. Version and language checked; onboarding and navigation completed.
  2. Built-in character opened; custom character created, edited, avatar changed, and new-chat avatar verified.
  3. Direct chat, solo roleplay, group chat, and group roleplay tested; resume, search, rename, and delete confirmed.
  4. History, actions, suggestions, branch, context, memory, relationship, journal, World, and report tested as entitled.
  5. Cloud provider tested without exposing a key; GGUF imported or downloaded, loaded, used, stopped, and unloaded.
  6. Piper or Supertonic downloaded with license, assigned, played, and stopped; Studio, emotions, lips, and avatar tested.
  7. Scene Director created/applied; languages, accessibility, and encrypted backup tested.
  8. Free/Pro, purchase/restore, or reviewer access observed without an unrequested real purchase.
  9. Report sent with steps, expected/actual result, version, language, and anonymized screenshots.
Settings and tester access with no visible secret
The checklist covers the full cycle without reviewer credentials.

30 · Reports and feedback

Send useful, safe evidence

Steps

  1. For one AI message, open History > Actions > Report this content, choose a category, and optionally comment.
  2. The report sends only the selected message and minimal metadata; the rest of the chat remains local.
  3. For an app bug, record version, language, Free/Pro status, screen, time, exact steps, expected result, and actual result.
  4. Attach a cropped screenshot only if it contains no API key, reviewer ID, real name, password, private backup, or restricted file.
  5. Include cloud/local model, GGUF size, and voice only when relevant. Replace sensitive values with [redacted].
  6. Send through the support channel listed on Google Play or the official website.
AI content report dialog
The report targets one selected message.

14 · Checklist

Before you send feedback

Steps

  1. onboarding and Home
  2. built-in and custom character
  3. solo chat and roleplay
  4. Group chat and Group roleplay
  5. one cloud model without exposing its key
  6. GGUF, load, streaming, and stop
  7. Piper or Android, then Supertonic
  8. Studio and built-in avatar
  9. authorized Pro VRM
  10. scene with local model
  11. accessibility and test-data backup
  12. sanitized report and Pro features

Verified on August 22, 2026 · Authentic screenshots · No reviewer data included.

↑ Back to contents