Selected website ingestion
You can now ingest selected websites into Skipper's local knowledge base for offline search and reference.
Changelog
Whenever there is a new functionality for Skipper being added or revised, it is recorded in the Changelog. Sometimes it is hard to describe in a simple summary what skipper can do. Hopefully through these snippets, Skipper capabilities can be explained with more detail.
Website capture and offline browsing
You can now use Skipper to capture selected websites for offline browsing whenever needed.
Created a Complete onboard custom model training
Skipper can train a genuinely modified LLM model onboard from a selected custom dataset, register it as a versioned runtime model, restore the complete Assistant, preserve traceability, and automatically clean…
Added Jetson performance mode selector
Easy switching between different GPU modes is now available through the Web UI.
Stable User Identity Implementation
User Authorizations and user Identity is now tightly bound to the session and the surrounding structure through a more organized architecture.
Secure Web UI Access
Added password-protected access and encrypted HTTPS connections for Skipper’s Web UI over both wired and onboard Wi-Fi networks. Additional safeguards help keep local access controlled while preserving Skipper’s fully offline…
Local Network Security Hardening
Tighter network configurations and security additions as well as encrypted verification layers.
Document Retrieval Presentation Improvement
Better presentation of ingested documents for Skipper as well as added relationships for Ingestions / Prompt presentation settings.
Added PDF Ingestion Improvements
The PDF ingestion system was revised so that documents are extracted, normalized, chunked, validated, previewed, stored, and indexed through one consistent production pipeline. (Better PDF PRocessing overall)
Added Temporal Conversation Memory
Skipper can now summarize previous conversations and include that summary in the discussion context.
Reject unrelated reference context for clear topics
Unrelated data to a deep seeded session topic is now being filtered out in long discussions.
Gate archived voice sessions with transcription quality scoring
Skipper can now separate and remember voice sessions better and with more accuracy.