~/hackyfeed $

Unstract is a tool designed to transform unstructured documents into structured JSON data using large language models (LLMs). It facilitates quick schema definition and document type adaptability through natural language prompts, enabling efficient extraction workflows for industries such as finance, healthcare, and compliance. Notable features include an intuitive Prompt Studio for schema creation, integration with multiple LLM providers, and straightforward deployment options via an API or ETL pipeline.

Taipy is a Python library designed for data scientists and machine learning engineers to build production-ready web applications that leverage data and AI. Its primary use case is to streamline the development of complex applications by automating tasks such as user interface generation, data integration, and pipeline orchestration while providing essential features like authentication and scheduling. Notable features include a comprehensive ecosystem with tools like Taipy Designer and Taipy Studio, as well as robust support for deployment and operational maintenance.

Supervision is a versatile toolkit for computer vision that facilitates data loading and real-time zone counting. It allows integration with any classification, detection, or segmentation model using connectors for popular libraries, streamlining the process of building applications around various model architectures. Notable features include its model-agnostic design, ease of installation, and comprehensive documentation for various use cases.

<⚡️> SuperAGI - A dev-first open source autonomous AI agent framework. Enabling developers to build, manage & run useful autonomous agents quickly and reliably.

STORM is an advanced LLM system designed to generate Wikipedia-like articles by conducting Internet-based research and forming outlines prior to writing. Its primary use case is aiding users in the pre-writing phase by creating structured content with citations, while the recently introduced Co-STORM enhances collaborative knowledge curation by allowing human-AI interaction for more aligned information retrieval. Notable features include support for various language models, integration with search engines, and a user-friendly demo interface for streamlined development and testing.

SASM - simple crossplatform IDE for NASM, MASM, GAS and FASM assembly languages

RFCrack is a Software Defined Radio (SDR) attack tool designed for testing RF communications across sub-GHz frequencies, targeting devices such as IoT systems, cars, and alarm setups. It features multiple attack functionalities including replay attacks, jamming, rolling code bypass, and the ability to compare live signals against recorded data, all of which can be executed with various modulation settings and frequency adjustments. RFCrack offers support for hardware like the Yardstick One and continuously evolves with new capabilities tailored to RF testing challenges.

Pelican is a static site generator developed in Python that enables users to create websites from plain text files using formats such as Markdown and reStructuredText. It generates entirely static websites that are easily hosted on any server, without the need for databases or server-side programming. Key features include support for multilingual articles, a customizable theming system, plugin extensibility, and fast rebuild times through content caching.

Mycli is a command-line client for MySQL-like databases that enhances the user experience with features such as auto-completion, syntax highlighting, and smart context-sensitive suggestions. It supports various databases, including MySQL, MariaDB, and TiDB, and offers notable capabilities such as fuzzy history search, query logging, and integration with tools like Pygments and fzf for better output handling. Users can also leverage advanced functionalities like querying LLMs and managing favorite queries with parameters.

NVIDIA Morpheus is an open AI application framework designed for cybersecurity developers, enabling rapid inspection of IP traffic across data centers with optimized AI capabilities. Its primary use case is to build customized pipelines for dynamic protection and real-time monitoring against cybersecurity threats, featuring pre-trained AI models, adaptive policies, and telemetry functionalities for effective threat detection and remediation. Notable features include a comprehensive command line interface for pipeline development, support for pre-built Docker containers, and extensive documentation for developers to extend its capabilities.

linkding is a self-hosted bookmark manager optimized for performance and ease of setup via Docker. It provides a clean user interface for organizing bookmarks with tags, supports bulk editing, and offers features like automatic website archiving, Markdown notes, and the ability to share bookmarks among users. Additionally, linkding supports SSO via OIDC, includes a REST API for third-party integrations, and features browser extensions for seamless access.

JARVIS is an artificial general intelligence (AGI) exploration tool that integrates large language models (LLMs) with various expert AI models to solve complex tasks through a four-stage process: task planning, model selection, task execution, and response generation. Notable features include the ability to utilize models from the Hugging Face Hub, support for Azure's OpenAI service, and a lightweight CLI mode for streamlined local deployment. Recent enhancements include the EasyTool for improved usability and TaskBench for benchmarking LLM automation capabilities.

InvokeAI is a versatile creative engine designed to facilitate the generation and refinement of visual media through advanced AI technologies. Its primary use case involves empowering artists and professionals to create, iterate, and manage workflows with a locally hosted web server and an intuitive user interface. Notable features include a Unified Canvas for collaborative creation, customizable node-based workflows, robust gallery management, and broad compatibility with various AI models, enhancing the flexibility and capability for users in their creative endeavors.

InboxInspector is a Streamlit application designed for cybersecurity analysis, enabling users to upload a CSV file of email addresses to query VirusTotal for domain reports associated with those addresses. The tool emphasizes user-friendliness with a straightforward interface for uploading files and viewing analysis results, while also providing features for downloading report outputs in JSON format. Notable aspects include dependency management via Poetry and the capability to operate within isolated environments, promoting a safe and efficient user experience.

GraphRAG is a data pipeline and transformation suite that leverages large language models (LLMs) to extract structured data from unstructured text, enhancing LLMs' reasoning capabilities concerning private data. Notable features include a methodology for integrating knowledge graph memory structures, support for prompt tuning to optimize output, and a command line interface for quick initial setup. The tool is particularly beneficial for users needing to refine LLM performance on narrative and private datasets.

FlipperZero-Subghz-DB is a comprehensive collection of over 13,700 Sub-GHz files specifically designed for the Flipper Zero device, facilitating research and educational exploration of Sub-GHz signals. The repository organizes these files into 721 folders, enhancing accessibility for users seeking to experiment with signal analysis and transmission. It serves as a centralized resource for the Flipper community, acknowledging contributions from various public sources.

FaceSwap is a deep learning tool designed for recognizing and swapping faces in images and videos, making it accessible for users without advanced technical knowledge. Its primary use case revolves around creating realistic face-swapping outputs for various applications, backed by models like Phaze-A and Villain. Notable features include a user-friendly GUI, comprehensive training capabilities, and support for community engagement through platforms like Discord and Patreon.

FaceFusion is an advanced face manipulation platform designed for technical users, enabling the execution of various commands related to face processing and job management. It supports multiple operational modes including headless and batch runs, along with extensive job handling capabilities such as creation, submission, and deletion of tasks. The tool requires installation via terminal or specialized installers for Windows and macOS, making it suitable for users with technical skills looking to leverage powerful face fusion capabilities.

ExLlamaV2 is an inference library designed for executing local large language models (LLMs) on modern consumer GPUs, providing high-performance capabilities through features like paged attention support and a consolidated dynamic generator API. It integrates seamlessly with the TabbyAPI backend to facilitate OpenAI-compatible local or remote inference, enabling advanced functionalities such as dynamic batching and smart prompt caching. While this project is currently archived, its enhancements focus on improving inference speed and efficiency, making it suitable for various LLM applications.

The Deepfake Offensive Toolkit (dot) enables real-time generation of deepfakes for virtual camera injection, primarily designed for penetration testing against identity verification and video conferencing systems. It features several methods, including face swapping and image animation, that function without the need for additional training, making it a versatile tool for security analysts, Red Team members, and biometrics researchers. The tool is intended for research and demonstration purposes, emphasizing the importance of compliance with applicable laws in its usage.

Deep-Live-Cam is a real-time deepfake tool enabling users to perform face swaps and create video deepfakes using just a single image. Its primary use case is in the AI-generated media industry, allowing artists to animate characters and produce engaging content seamlessly. Notable features include an easy three-step process for live swapping, mouth masking for accurate movement retention, and support for multiple faces on different subjects simultaneously, along with built-in safeguards to prevent unethical or inappropriate use.

ComfyUI-Manager is an extension that enhances the functionality of ComfyUI by providing management tools for installing, removing, enabling, and disabling custom nodes. It features a centralized hub for easy access to various node information and includes recent security improvements, support for UV, and integration with the ComfyUI registry. This tool streamlines user interaction with ComfyUI, making it more efficient and secure.

Codon is a high-performance Python implementation designed for static, ahead-of-time compilation, achieving significant speed improvements (10-100x) over traditional Python, often matching or surpassing the performance of C/C++. It enables native multithreading, seamless hardware support, and interoperability with existing Python libraries while maintaining a syntax and structure that closely resembles CPython, thus minimizing the learning curve for users. Notable features include a comprehensive optimization framework and tools for compiling Python code to native machine code or executable files.

BabyAGI is an experimental framework designed for creating self-building autonomous agents by facilitating task planning and execution through function management. Its notable features include a graph-based structure for tracking function dependencies and imports, a dashboard for function management, and comprehensive logging capabilities, making it a developer's playground for exploring autonomous agent development. While the project emphasizes innovation and discussion rather than production use, it provides a robust environment for experimentation with function registration and metadata management.

The Splunk Attack Data repository provides a collection of curated datasets from various cyber attack simulations, facilitating the development and testing of detections in Splunk environments. Key features include the ability to replay these datasets into streaming pipelines for real-time validation of detection mechanisms, as well as integration with Splunk's Security Content for streamlined analysis. This tool helps security professionals avoid the complexities of setting up attack simulations from scratch, thereby accelerating their detection capabilities.