v 0.10.1

[Ghassan-elsman]

Initialization and Installation Pipeline

• Re-enabled the Forensic Dependencies Installer to automatically manage pure-Python dissect libraries and AI SDKs (OpenAI, Anthropic, Gemini) on Python 3.12.
• Implemented a self-healing React build pipeline for the Timeline and EYE Assistant interfaces to resolve deployment inconsistencies.
• Added automatic environment detection to ensure locally installed Node.js and NPM are correctly utilized during the build process.
• Integrated a mandatory post-build patching step to inject browser polyfills and layout fixes required for internal QtWebEngine stability.

Backend Resilience and Local Model Support

• Hardened the LM Studio backend to detect and handle HTTP 400 errors specifically caused by servers running without a loaded model.
• Enhanced the Gemini backend to support both legacy and modern Google GenAI SDK model discovery properties, ensuring compatibility with version 1.0+ of the SDK.
• Implemented automatic model detection for LM Studio and local API backends to resolve connection errors when specific model names are not manually provided.
• Enhanced pre-flight connectivity checks to distinguish between offline servers and unconfigured model states in local LLM environments.
• Added specific troubleshooting instructions for local model servers, guiding users to load models into memory within the server application.
• Implemented schema-aware query fallbacks in the forensic engine to handle variations in artifact database structures and missing columns without process termination.
• Unified credential management across UI components to ensure real-time synchronization of API keys and server endpoints.
• Corrected internal schema documentation naming conventions and system prompt references to improve RAG accuracy and SQL generation.

UI and Platform Stability

• Resolved the Eye Assistant launch failure by moving AA_ShareOpenGLContexts initialization to the main application entry point, ensuring Qt OpenGL shared context stability across all windows.
• Refined the Onboarding Wizard to persist and pre-populate existing configuration data, preventing redundant setup steps during reconfiguration.
• Improved the Onboarding Wizard diagnostic suite to provide real-time remediation steps for missing SDKs or corrupt build artifacts.
• Integrated context-aware troubleshooting advice for local network and cloud API configurations within the setup interface.
• Standardized multi-window lifecycle management to prevent application crashes during asynchronous AI generations and tool executions.
• Updated the system diagnostics tool to verify Python dependencies, SDK integrity, and frontend React build status prior to launch.

Full Changelog: 0.10.0...0.10.1

---

This discussion was created from the release v 0.10.1.