Skip to content

v0.3.0-alpha.2

Pre-release
Pre-release

Choose a tag to compare

View all tags
@Baronco Baronco released this 28 Dec 01:23
· 63 commits to dev since this release
v0.3.0-alpha.2
0985a20

v0.3.0-alpha.2 Image Embedding in DOCX & Enhanced Agent Behavior 馃殌

Added support for embedding images directly into generated Word documents from chat uploads. Updated system prompt for more autonomous document generation and improved chat_context tool to retrieve image metadata.

Highlights & Objectives 馃搶

  • What changed
    • Image Embedding: New capability to include images uploaded to the chat in generated DOCX files. Images are embedded seamlessly, preserving order and context.
    • System Prompt Update: Enhanced FileGenAgent prompt to reduce clarification requests, allowing the model to make assumptions and proceed autonomously for better user experience.
    • chat_context Tool: Now retrieves image IDs and metadata from chat messages, enabling access to uploaded images for document generation.
  • Why
    • Enable richer, visual content in documents without external dependencies.
    • Improve efficiency by minimizing back-and-forth interactions, letting the AI infer and generate based on context.
    • Streamline image handling for more dynamic and context-aware file creation.
  • Compatibility
    • Requires Open Web UI v0.6.42+ (due to knowledge API changes).
    • For OWUI < v0.6.42 use GenFilesMCP <= 0.2.2.
    • MCPO: alpha is currently not compatible with MCPO; compatibility depends on PR open-webui/mcpo#273 (adds per-session bearer token header support). Goal: full MCPO compatibility once that PR is merged.
Assets 2
Loading