DeepTutor 1.4.x Roadmap
With v1.4.0 shipping as the next minor line, here's what we're aiming at across 1.4.x. Replies, +1s, and PRs are very welcome — scope and ordering will adjust based on community feedback.
🎯 Themes
- MCP tool — Promote MCP from a TutorBot-only integration to a first-class tool source for every main capability (Chat / Deep Solve / Deep Research / Co-Writer / Book), with a Settings page to register servers and CLI parity.
- Voice I/O + Image / Video generation — Voice input/output in the Chat composer and TutorBot channels (pluggable STT/TTS), plus provider-agnostic
image_generation and video_generation tools callable from any capability.
- TutorBot update — Catch up with the latest nanobot capabilities and align our agent loop, Soul / Skill / Heartbeat surfaces, and sub-agent contract with upstream.
- LightRAG integration — Integrate our sibling HKUDS project LightRAG into DeepTutor's RAG layer as a selectable backend alongside the current LlamaIndex pipeline.
- Multi-user - Bug fixing and function refactoring. (This function is not the main target of 1.4.x.)
🤝 We always welcome PRs
DeepTutor warmly welcomes any kind of adaptation toward the open-source community — bridges to other open-source projects, runtimes, frameworks, deployment targets, or ecosystems. If it helps DeepTutor meet users where they already are, we want it.
The most consistently useful contributions are provider adapters. If a service you care about doesn't work out-of-the-box, a PR that wires it up is gold:
- LLM providers — new chat-completion / responses-API endpoints, thinking-model routing, tool-call quirks.
- Embedding providers — new embedding backends and dimensional / signature handling for the KB layer.
- Search providers — additional web / academic search backends behind the existing search abstraction.
Small focused PRs that just wire up one provider are very welcome — they ship fast and help everyone.
See CONTRIBUTING.md for setup and coding standards. For bigger design discussions (memory graph schema, MCP UX, LightRAG benchmarks), open a Discussion and we'll link it back here.
DeepTutor 1.4.x Roadmap
With v1.4.0 shipping as the next minor line, here's what we're aiming at across 1.4.x. Replies, +1s, and PRs are very welcome — scope and ordering will adjust based on community feedback.
🎯 Themes
image_generationandvideo_generationtools callable from any capability.🤝 We always welcome PRs
DeepTutor warmly welcomes any kind of adaptation toward the open-source community — bridges to other open-source projects, runtimes, frameworks, deployment targets, or ecosystems. If it helps DeepTutor meet users where they already are, we want it.
The most consistently useful contributions are provider adapters. If a service you care about doesn't work out-of-the-box, a PR that wires it up is gold:
Small focused PRs that just wire up one provider are very welcome — they ship fast and help everyone.
See
CONTRIBUTING.mdfor setup and coding standards. For bigger design discussions (memory graph schema, MCP UX, LightRAG benchmarks), open a Discussion and we'll link it back here.