Open Source Digital Life Companion Framework
简体中文 · English
Quick Start · ZIP Users · Deploy · Commands · Docs
Waveary is an open source framework for long-term digital companionship.
It gives any compatible model a continuity layer for:
- long-term memory
- relationship growth
- life timeline awareness
- emotional continuity
- voice interaction
Waveary is not trying to be just another chatbot skin.
It is trying to be the system layer that helps an AI remember, understand, grow, and stay with a user over time.
Current usable surface:
- browser-based chat UI
- provider setup and model discovery
- persistent local chat sessions
- memory / relationship / timeline runtime signals
- dedicated voice routing foundation
Current project status:
- suitable for developers, builders, and early testers
- not yet a one-click packaged consumer app
- easiest working path today is local run or self-hosted Node deployment
If you just want to run Waveary locally, do this:
- Node.js 20+
npm 10+(usually comes with Node.js)
Check your versions:
node -v
npm -vUse Git:
git clone https://github.com/K2st0r/Waveary.git
cd WavearyOr download the ZIP from GitHub, then extract it and open the extracted folder in your terminal.
npm installnpm run web:devWhen startup succeeds, open:
http://127.0.0.1:4173
After the page opens:
- Go to the console page.
- Choose a provider.
- Fill in
Base URLandAPI Key. - Click to fetch models.
- Choose a model.
- Start chatting.
You can also use the CLI helper:
npm run setup:providerThat command will:
- list known providers
- ask for your API key
- fetch available models
- save the selected config locally
If you downloaded the project as a ZIP and do not use Git, follow these steps exactly:
- Download the project ZIP from GitHub.
- Right-click the ZIP and choose
Extract All. - Open the extracted
Wavearyfolder. - Click the folder path bar, type
powershell, and press Enter. - Run:
npm install
npm run web:dev- Open
http://127.0.0.1:4173in your browser.
- Download and extract the ZIP.
- Open Terminal.
cdinto the extracted folder.- Run:
npm install
npm run web:dev- Open
http://127.0.0.1:4173.
The simplest supported deployment path today is:
- Linux server
- Node.js 20+
npm run web:preview- reverse proxy through Nginx or Caddy
git clone https://github.com/K2st0r/Waveary.git
cd Waveary
npm install
npm run web:previewBy default, Waveary preview runs on:
http://127.0.0.1:4173
For public access, put Nginx or Caddy in front of it and reverse proxy to 127.0.0.1:4173.
Detailed step-by-step deployment instructions are here:
# Start local development
npm run web:dev
# Build the web app
npm run web:build
# Run local production preview
npm run web:preview
# Full test run
npm run test
# Interactive provider setup
npm run setup:provider
# Verify saved provider config
npm run verify:provider
# List models from current provider config
npm run models:provider
# Reset local test chat memory only
npm run reset:test-memoryWaveary stores local runtime data in the repository root under:
.waveary/
Common files include:
.waveary/provider-config.json.waveary/chat-sessions.json.waveary/chat-sessions.db
If you move the project to another machine and want to keep your local state, back up the .waveary/ folder too.
Install Node.js first, then reopen your terminal.
Try:
npm cache verify
npm installCheck whether the dev server is running and whether port 4173 is already in use.
Run:
npm run verify:providerThis checks whether your saved provider config can actually:
- list models
- select a usable model
- finish one real chat turn
Run:
npm run reset:test-memorywaveary/
waveary-core
waveary-memory
waveary-voice
waveary-web
waveary-dataset
docs
Module roles:
waveary-core: runtime orchestration and provider abstractionwaveary-memory: memory extraction, storage, retrievalwaveary-voice: voice routing and voice runtime layerwaveary-web: official web surfacewaveary-dataset: markdown-first companion soul and conversation rules
- Chinese README
- Deployment Guide
- 中文部署指南
- Project State
- Vision
- Architecture
- Product Invariants
- Product Preferences
- Web Surface
Issues, architecture discussion, feature proposals, and focused pull requests are welcome.
Basic flow:
- Fork the repository.
- Create a branch.
- Make one focused change.
- Open a pull request.
This project is released under the MIT License.