askimo Files

Summary

This release focused on architectural modernization and the integration of advanced AI capabilities. Core systems were updated by migrating the database layer to SQLDelight and refactoring caching and logging mechanisms to utilize modern backends (cache4k and Kermit). Functionality was extended by introducing local Large Language Model (LLM) and Whisper voice integrations, alongside enhancements to the suggestion and memory management systems. Various components also received structural updates, including standardized time handling and generalized API timeout controls, improving system robustness.

What's Changed

• Migrated database interactions from legacy methods to SQLDelight, improving type safety and structured query access (#742)
• Replaced the shared caching implementation with cache4k, standardizing the caching module across the application (#744)
• Updated Docker provider settings to support a generalized migration path for legacy configurations (#741)
• Switched the logging system to Kermit logging backend, formalizing the event logging architecture and removing legacy event handlers (#751)
• Standardized time management throughout the codebase by adopting Instant and Clock interfaces, ensuring consistent time zone and time arithmetic handling (#752)
• Implemented generalized model request timeout handling, applying a consistent time limit structure across all external API calls (#746)
• Enhanced the core system by improving exception mapping and expanding detailed analytics reporting mechanisms (#760)
• Added support for local LLM engine integration, enabling the execution and management of models running client-side (#762)
• Integrated Whisper model support for voice processing, including necessary model download and lifecycle management features (#756)
• Implemented structured management for Whisper model downloads, including the ability to cancel ongoing downloads, improving user control (#767)
• Developed typing suggestion features for inline autocomplete, enhancing the user input experience based on learned patterns (#759)
• Enhanced the chat suggestion module with improved matching algorithms and a dedicated mechanism for managing usage history (#761)
• Updated core functionality to enhance retention tracking, tool handling state management, and to manage the dedicated timeout state for tool approvals (#754, #755)
• Introduced automated generation of third-party license reports, streamlining the build and compliance process (#757)
• Improved the summarization and memory management architecture, providing more efficient handling of conversation context (#765)
• Added the capability to load embedding models from a dedicated model catalog, streamlining the initialization process for the open-router module (#763)
• Updated UI components by adjusting the navigation sidebar's hover height and implementing logic to hide the header when no session is active (#750)
• Added links to community forums and refined the error reporting system within the user interface (#753)

Contributors

• @haiphucnguyen
• @jeonjw85
• @AcynD
• @larry-ca
• @totodev90
• @toktokbr
• @mikaniisan
• @huynguyen20011

Full Changelog: https://github.com/askimo-ai/askimo/compare/v1.5.3...v1.5.4