All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
- Add llm.md
- Frontend
- serialization.ts
- Platform
- fully support k8s
- Frontend
- add middleware for node check
- better UI, UX,
-
Frontend
- Fully switch to SvelteFlow
- Refractor to logic, only Node is SSOT, signals to edge update
-
Backend
- change process filename as workflow
- Frontend
- change to using redux for config
- Release
- Backend
- supervisord + daily restart by cron
- Backend
- k8s hold at yourdomain.com/backend
- tls for backend
- Frontend
- k8s hold at yourdomain.com
- tls for frontend
- backend
- k8s support
- compatible with docker compose + k8s
- Release Logic
- add github release and upload with ver
-
Backend
- llm support ollama any
-
ollama
- mount .ollama to prevent need download again
-
example
- better trpg play prompt
- Backend
- support subgraph
-
Backend
- support subgraph (but run "root" only)
-
Doc
- change to new UI
- Frontned
- Chage to typescript
- make redux as SSOT to update reactflow
- change to @xyflow/react
- Workflow decouple langchain, TODO: clip move to llm.py also.
-
General
- New node type: INFO
- New node format: ext, this can extend for dynamic parsing
-
Frontend
- refractor to 2 repo,
src/Graphand repo wrapper - new run, better UI/UX
- use tailwind
- support INFO node
- refractor to 2 repo,
-
backend:
- using FastAPI
- refractor as llm.py and workflow.py
- use workspace/undefined, better for multi user is login with nginx auth with name
- support multi user if using nginx, example at
nginx/nginx.conf - support change port by env var
-
ollama
- support change port
- support NV GPU
- Enable Japanese Input in Node Components by @hangingma
- upload files