SeedOSS Cloud
764 subscribers
1.07K photos
18 files
43 links
Where images and information is stored 🙃.
Download Telegram
Screenshots of Paperless-ngx
Features of LastChat

🤖 Multi-Provider Support: Connect to any AI provider with built-in presets or add your own custom endpoint.
🧠 RAG Memory: Assistants remember details from past conversations using a local embedding-based memory system.
🎭 Multiple Personas: Create and manage unlimited custom assistants with tags, import/export, and fast switching.
🐍 Python and JS Engines: Run Python via Chaquopy and JavaScript via QuickJS directly inside the app.
🌐 Web Search and MCP: Fetch real-time information via integrated web search or connect to any MCP server.
🖼️ Image Generation: Dedicated interface for generating images through supported models.
🔒 Local-First Storage: Chat history and vector memory stay on your device with optional WebDAV backup.
🎨 Material You 3 Expressive: Built with the latest M3E design spec and full Markdown, LaTeX, and code highlighting support.
1
Screenshots of LastChat
OhMyCaptcha — Python/Website
A self-hostable captcha solver exposing a YesCaptcha-compatible API for flow2api and similar integrations. Combines FastAPI, Playwright for reCAPTCHA v3 solving, and OpenAI-compatible multimodal models for image captcha recognition, with in-memory async task management.
Why it's amazing
🔗 Get it: Website
📁 Source Code

❤️ Sincerely: @SeedOSS
Features of OhMyCaptcha

🔄 YesCaptcha-Compatible API: Drop-in replacement for createTask, getTaskResult, and getBalance endpoints used by flow2api and routing layers.
🤖 reCAPTCHA v3 Solving: Handles RecaptchaV3TaskProxyless and multiple M1/S7/S9 variants using Playwright and headless Chromium.
🖼 Image Captcha Recognition: Processes ImageToTextTask using OpenAI-compatible multimodal models with Argus-inspired normalization to 1440×900 coordinates.
🏠 Self-Hosted Deployment: Designed for internal workflows with full control over browser environment, model routing, and API endpoints.
⚡️ In-Memory Task Manager: Lightweight async background execution with configurable TTL and retry logic, no external database required.
🌐 Multimodal Model Support: Integrates with hosted or self-hosted OpenAI-compatible gateways for flexible model selection.
📚 Dual Documentation: Complete English and Simplified Chinese guides published via GitHub Pages with deployment instructions for Render and Hugging Face Spaces.
🐳 Render-Ready Docker: Includes Dockerfile and render.yaml for one-click deployment to cloud platforms.