The biggest release yet! eVera transforms from a basic assistant into a full Jarvis-like AI copilot with multi-model intelligence, personal knowledge base, Chrome extension, and autonomous web automation.
| File | Description |
|---|---|
| Vera Setup 0.9.0.exe | Windows installer (includes everything — no Python/Node needed) |
- 9 providers: OpenAI, Anthropic (Claude), Google Gemini, Groq, DeepSeek, Mistral, Together AI, Perplexity, Ollama
- Intelligent auto-routing: Code → DeepSeek, Creative → Claude, Fast → Groq, Vision → GPT-4o, Web Search → Perplexity
- Model selector in the Web UI header — pick any model or let eVera choose
- Provider health checks — see which providers are online
- Per-model metadata: context window, vision support, tool support, cost tracking, speed tier
- Upload documents: PDF, DOCX, TXT, Markdown, CSV (up to 50MB)
- Smart chunking: 512-token chunks with 64-token overlap for optimal retrieval
- Vector search: sentence-transformers + FAISS for fast similarity search
- RAG answers with citations: Answers reference which document and chunk the info came from
- Persistent storage: Documents survive server restarts
- Text selection toolbar: Select text on any webpage → floating toolbar with 5 AI actions
- Actions: 📝 Summarize, 🌐 Translate, 💡 Explain, ✏️ Fix Grammar, 🤖 Ask eVera
- Right-click context menus for all actions
- Side Panel chat: Full chat interface in Chrome's sidebar
- Settings page: Configure server URL, default model, toggle overlay
- Glassmorphism UI matching the desktop app
- Natural language → browser automation: "Search Google for Python tutorials" → plans and executes multi-step sequences
- LLM-powered task planner: Decomposes intent into navigate/click/type/scroll steps
- Screenshot verification: Takes screenshots after each step to verify success
- Auto-retry & re-planning: Falls back to alternative selectors or re-plans on failure
- Web scraper: Extract structured data from any page with a JSON schema
- Backend now bundles correctly with PyInstaller (20+ hidden imports added)
.envresolution works in frozen exe mode (_MEIPASS+%APPDATA%)- No more duplicate browser window opening in Electron
waitForBackend()properly rejects on timeout with error dialog- Data directories auto-created on first run
- One-click
scripts/setup.ps1setup script
| Metric | Value |
|---|---|
| Files changed | 38 |
| Lines added | 8,294 |
| New files | 24 |
| LLM providers | 9 |
| Models supported | 30+ |
| Agents | 20 |
| Extension files | 10 |
| Test coverage | 21/21 passing |
git clone https://github.com/srpatcha/eVera.git
cd eVera
powershell -ExecutionPolicy Bypass -File scripts/setup.ps1
python main.py --mode server
# Open http://localhost:8000 in Chromecd electron
npm startEdit .env and add your provider keys:
VERA_LLM_OPENAI_API_KEY=sk-...
VERA_LLM_ANTHROPIC_API_KEY=sk-ant-...
VERA_LLM_GROQ_API_KEY=gsk_...
VERA_LLM_DEEPSEEK_API_KEY=sk-...
- Open
chrome://extensions - Enable Developer Mode
- Click "Load unpacked" → select
extension/folder
| Endpoint | Description |
|---|---|
GET /models |
List 30+ models with availability status |
POST /models/select |
Auto-pick best model for a task type |
GET /models/health |
Provider health check |
POST /knowledge/upload |
Upload document to knowledge base |
GET /knowledge/documents |
List indexed documents |
POST /knowledge/query |
RAG query with citations |
POST /extension/summarize |
Summarize text |
POST /extension/translate |
Translate text |
POST /extension/explain |
Explain text |
POST /extension/grammar |
Fix grammar |
POST /extension/rewrite |
Rewrite text |
POST /automation/plan |
Plan browser automation |
POST /automation/execute |
Execute browser automation |
POST /automation/scrape |
Scrape structured data |
Full Changelog: https://github.com/srpatcha/eVera/compare/v0.6.0...v0.9.0