9 dépôts
Techniques for keeping decrypted data exclusively in volatile memory.
Distinct from In-Memory Data Stores: Distinct from general in-memory stores: focuses on the security-driven isolation of decrypted secrets from persistent storage.
Explore 9 awesome GitHub repositories matching data & databases · Memory-Isolated Decryption. Refine with filters or upvote what's useful.
Gluetun is a Docker VPN client gateway that routes network traffic from other containers through secure WireGuard or OpenVPN tunnels. It serves as a network gateway to mask origin IP addresses and provides a firewall kill switch that blocks all outbound traffic unless it is destined for the active VPN server. The project supports multiple commercial VPN providers and can maintain several simultaneous connections using custom configuration files. It includes an integrated proxy server stack hosting SOCKS5, HTTP, and Shadowsocks servers to tunnel TCP and UDP traffic. Security is managed throug
Decrypts sensitive VPN provider keys into volatile memory at runtime to avoid writing plaintext secrets to disk.
This project is a comprehensive zero-knowledge security suite designed for enterprise credential management, secrets orchestration, and password management. It provides a secure, end-to-end encrypted vault that allows users to store, synchronize, and manage sensitive information, including passwords, passkeys, and infrastructure secrets, across desktop, mobile, and browser environments. The platform distinguishes itself through a strict zero-knowledge architecture where all encryption and decryption occur locally on the client, ensuring that plaintext data remains inaccessible to the server.
Ensures sensitive data is never written to persistent disk storage by storing decrypted vault information exclusively in volatile system memory.
Android-Debug-Database is a specialized utility for extracting, inspecting, and editing mobile data on Android devices. It serves as a database debugger and SQLite inspector that provides a web-based interface for managing database records and shared preference key-value stores. The project distinguishes itself by supporting encrypted database decryption via provided passwords and the ability to map and inspect volatile in-memory databases. It also includes a data export tool that transfers database files from the private application directory to a local machine for external analysis. The to
Decrypts database files in memory using provided passwords before displaying content in the inspector.
The term 'on-the-fly' appears in the feature name but the candidate is unrelated to script injection.
Paperless-ng est un système de gestion documentaire auto-hébergé conçu pour archiver la paperasse physique sous forme de fichiers numériques consultables. Il fonctionne comme un serveur privé pour scanner, indexer et organiser une bibliothèque numérique de documents via une interface web. Le système agit comme une archive de fichiers chiffrée, utilisant un backend de protection de la vie privée pour sécuriser les documents stockés. Il fournit un déchiffrement automatique à la volée pendant le processus de téléchargement pour garantir que les archives restent protégées pendant le stockage. La plateforme intègre la reconnaissance optique de caractères (OCR) pour convertir les images scannées et les PDF en texte consultable pour l'indexation en base de données. Elle prend également en charge l'ingestion automatique en surveillant les répertoires locaux ou réseau pour les nouveaux téléchargements.
Provides real-time decryption of stored documents during the request cycle to maintain storage security.
KeePassX is a cross-platform password vault application and database manager that stores usernames and passwords in an encrypted file. It utilizes the KeePass database format to secure records using a master password or an optional key file. The application includes an automatic form filler that populates login fields in external applications and a secure password generator for creating high-entropy randomized passwords. The system provides tools for organizing credentials through custom metadata and grouping, alongside utilities for searching records and importing or exporting data.
Keeps decrypted records in volatile system memory to allow fast access without repeated decryption.
tiny-llm est un moteur d'inférence de grands modèles de langage et une implémentation de modèle transformer. Il sert de runtime pour modèles quantifiés et de gestionnaire de cache clé-valeur paginé, fournissant une pile d'inférence spécialisée optimisée pour Apple Silicon. Le système se distingue par des techniques d'exécution à haut débit, incluant le batching continu et l'attention paginée. Il utilise un système de mémoire paginée pour éliminer la fragmentation lors de la génération de jetons et emploie une déquantification à la volée des poids compressés pour réduire l'empreinte mémoire lors de la multiplication matricielle. Le projet couvre un large éventail de capacités d'architecture de modèle et de performance, telles que le routage par mélange d'experts (MoE), l'attention par groupes de requêtes (GQA) et l'attention flash. Il inclut la prise en charge d'une logique de décodage avancée, incluant le décodage glouton (greedy) et l'échantillonnage via des méthodes de température, top-k et top-p. L'implémentation est écrite en Python et inclut des noyaux bas niveau personnalisés pour accélérer le traitement des tenseurs sur le matériel.
Performs on-the-fly recovery of original precision from compressed weights during model inference.
git-secret is a command line tool and Bash encryption utility used to manage sensitive configuration files and passwords within Git repositories. It enables version controlled secret storage by encrypting files with GPG public keys, ensuring that sensitive data can be committed to a repository without exposing plaintext. The tool utilizes a PGP encryption workflow to control access through a managed keyring of authorized public keys. This allows for the granting and revocation of decryption permissions for specific users. To prevent accidental data leaks, it automatically integrates with Git
Decrypts files into memory or streams to allow viewing and diffing without writing plaintext to disk.
ComfyUI-GGUF is a memory optimizer and model loader for ComfyUI that enables the execution of large transformer-based generative models using quantized weights. It provides a system for loading GGUF formatted weights within a node-based diffusion interface to reduce GPU memory consumption. The project includes a quantization tool for converting standard model checkpoints into compressed binary formats and a tensor fixer to restore missing keys and correct architectures in binary model files. These utilities ensure that compressed models remain functional during inference on hardware with limi
Enables high-performance inference by recovering weight precision on-the-fly from compressed memory footprints.