llmConnect is a local-first desktop workspace for browser-based AI chats and encrypted API profiles. It uses PyQt6 to keep browser sessions isolated, extract structured chat history, and organize everything in a dashboard that stays on your machine.
AI usage tends to fragment across browser tabs, sessions, and API accounts. llmConnect brings that work into one place so you can:
- keep browser sessions separated,
- save conversations locally,
- organize chats by category and project,
- manage API profiles securely,
- and review structured chat history later instead of losing it in the browser.
- Opens separate browser tabs for supported providers.
- Injects lightweight browser tools for persona prompts and extraction.
- Stores extracted chats in SQLite.
- Stores API profiles encrypted on disk.
- Categorizes chats into titles, categories, and projects.
- Renders recent chats in a thread-style dashboard.
Start with .docs/README.md.
Recommended reading:
main.py: Application entry point.engine/: Storage, classification, and local integration logic.models/: Browser/provider selector definitions.ui/: Main window, browser tabs, and dashboard screens.skills/: Prompt and role markdown files for browser and API workflows.
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python main.py- Extraction depends on the current browser DOM of each provider.
- Browser tools can stop working if a site changes its UI or script restrictions.
- Validation and categorization depend on provider availability and network access.
See .docs/limitations-and-roadmap.md for the detailed discussion and future direction.
