Cerbrec Graphbook
Move Operations from Reaction to Resilience
Graphbook is a deep learning framework with a visual interface, transforming cumbersome model development into a user-friendly experience. With Graphbook, you can edit and debug deep learning models as a drag-and-drop data orchestrator inside your models. Cerbrec Graphbook is the AI development platform for everyone to create bespoke AI solutions, by natural language queries or drag-and-drop.
What To Build
Decision Intelligence for Builders
What To Build is an AI-powered platform designed to assist developers and open-source enthusiasts. It helps users generate project ideas, find similar repositories, and create visual plans for building projects. The platform offers a suite of tools to enhance GitHub workflow, including features for transforming ideas into structured GitHub repositories, analyzing existing repositories, generating professional READMEs, visualizing repositories, finding suitable open-source projects for contribution, and comparing GitHub developers based on their activity and contributions.
Codewise
Translate business requirements into code
Codewise is an AI-powered SaaS tool designed to translate non-technical business requirements into functional code. It centralizes your favorite tools and data in one place, aiming to streamline the software development process.
Pencil – Design on canvas. Land in code.
Design on canvas. Land in code.
Pencil is an agent-driven Multi-Channel Platform (MCP) canvas designed to fundamentally increase engineering speed by integrating design directly into the preferred Integrated Development Environment (IDE), such as VSCode and Cursor. It promotes 'Design right where you code' by utilizing an open design format that lives within the codebase. Pencil offers an infinite design canvas, AI multiplayer capabilities for parallel screen generation, precision tools for prompting, and a vector-to-code workflow to ship pixel-perfect, production-ready applications where code and design are always aligned and versioned with Git.
Streamdown
A markdown renderer designed for streaming content from AI models.
Streamdown is a drop-in replacement for react-markdown, specifically designed for AI-powered streaming. It was built to enable the streaming of safe and perfectly formatted Markdown content without complex handling, powering components like the AI Elements Response component. It can be installed and used as a standalone package, offering features like built-in typography styles, GitHub Flavored Markdown support, interactive code blocks with Shiki highlighting, mathematical expressions via LaTeX and KaTeX, interactive Mermaid diagrams, and robust security hardening against prompt injection.
metaflow.org
A framework for real-life ML, AI, and data science
Metaflow is an open-source framework originally developed at Netflix for building and managing real-life ML, AI, and data science projects. It allows data scientists and ML/AI engineers to develop, debug, and deploy workflows in plain Python, leveraging cloud resources for scalable compute and data access. Metaflow integrates seamlessly with existing infrastructure, security, and data governance policies, supporting deployment on AWS, Azure, Google Cloud, and Kubernetes.
Silo Team
The fastest way to onboard and offboard people
Silo Team is an onboarding platform that helps managers onboard developers faster with AI. It automates everything, integrates with your tools, and provides full visibility, eliminating checklists and delays. It also supports offboarding.
Rerun
Debug and visualize computer vision and robotics data with minimal code.
Rerun is an SDK for logging computer vision and robotics data paired with a visualizer for exploring that data over time. It lets you debug and understand the internal state and data of your systems with minimal code. It’s built in Rust to run fast everywhere. Rerun provides multimodal log handling and visualization for spatial and embodied AI. It offers managed infrastructure to ingest, store, analyze, and stream data at scale with built-in visual debugging. It is designed to be fast, flexible, and easy to use.
WorkWeave
Engineering intelligence for the AI era.
WorkWeave turns any engineer into a 10x engineer. It connects to everywhere work happens, feeds the data to its AI performance enhancement platform, and turns it into personalized insights and actionable feedback. Weave gives you the data you need to optimize your engineering team, with intelligent algorithms that deeply understand your work.
SvectorDB
From prototype to production in a few lines of code
SvectorDB is a serverless vector database built from the ground up for AWS. It offers cost-effective vector search, allowing users to pay only for what they use and optimize their cloud spend. It is designed to be high-performance and up to 20x cheaper than alternatives, focusing on enabling users to concentrate on their product rather than database management.
theterminal.app
A hands-on learning platform for DevOps and SRE engineers.
TheTerminal is a hands-on learning platform for DevOps and SRE engineers. It offers real-world challenges, AI-driven performance analysis, and tools like an embedded terminal and code editor to improve skills in areas like Containers, Networking, Security, Infra as Code, CI/CD, and Scripting.
determined.ai
HPE machine learning development environment with automated hyperparameter optimization
Determined AI is an HPE machine learning development environment that provides automated hyperparameter optimization, distributed training, cluster management, and enterprise-grade cloud solutions. The documentation provides quick access to information, tutorials, quickstarts, user guides, and reference material.
captum.ai
Model Interpretability for PyTorch
Captum is a PyTorch library for model interpretability. It provides tools and techniques to understand and attribute the predictions of PyTorch models across various modalities, including vision and text. Captum supports most PyTorch models and allows for easy implementation and benchmarking of new interpretability algorithms.
Nitrode
Technology x Finance: Smart, Connected, Proven
Nitrode is an AI game engine designed to empower developers to quickly create playable 3D games in a matter of hours. It is an AI-native engine that facilitates game design, scripting, and deployment by allowing users to create mechanics through connecting nodes, generate assets with AI, and build games without boilerplate code. Nitrode aims to help users realize their dream games by providing smart building tools and AI-powered assistance.
mcp-use
Build and deploy MCP apps and servers
mcp-use is an open-source platform that provides devtools and cloud infrastructure to help development teams quickly build and deploy custom AI agents using MCP (Model Context Protocol) servers. It offers an SDK for creating AI agents, a registry for discovering community-built servers, and a managed gateway for routing, authenticating, and load-balancing MCP servers. The platform aims to simplify the process of building, discovering, and deploying MCP servers and AI agents, supporting various deployment environments including cloud, local VMs, and third-party services.
New API
One API endpoint for all your AI model providers
New API is a unified AI model hub that aggregates multiple LLM providers into a single API endpoint. It handles key management, usage tracking, billing, and load balancing for OpenAI, Anthropic, Google, Mistral, and dozens of other AI providers through one consistent interface.
Progressier
PWA Made Easy
Progressier is a toolkit designed for web developers and "vibe coders" to effortlessly build fully-featured Progressive Web Apps (PWAs). It enables universal installation across various devices (iPhone, iPad, Android, Mac, PC) and browsers, and provides robust push notification capabilities. The platform simplifies PWA development by offering no-code solutions for integrating installation and push notifications into existing web apps, managing app manifests, generating necessary assets like splash screens and maskable icons, and providing tools for PWA promotion, analytics, and even programmatic PWA creation. It also includes features like an app reviews manager and integrations with numerous no-code builders and JavaScript frameworks.
chainlit.io
Build AI applications
Chainlit is a framework and platform designed to help developers and enterprises build, evaluate, and improve conversational AI applications. It provides tools for creating reliable AI systems, offering observability and analytics for LLM apps. Chainlit supports various deployment options, including standalone web apps, embedded copilots, FastAPI servers, and custom React or Slack/Discord/Teams bots. It also offers authentication and customization features, along with integrations with popular frameworks and LLM providers.
Cencurity
Enterprise security gateway for LLM agents
Cencurity is an enterprise-grade security gateway designed to proxy LLM and agent traffic. It functions by detecting, masking, and blocking sensitive data, PII, and risky code patterns in both requests and responses. The platform provides a centralized dashboard for real-time monitoring and recording of all interactions as audit logs. It is designed to be compatible with any LLM agent or IDE without requiring code rewrites, ensuring safe and governed AI integration for development teams.
cubic
AI code reviewer that merges PRs 48% faster
cubic is an AI code reviewer for complex codebases. It’s used by top engineering teams at n8n, Cal.com, Resend, and more. Code review is a major bottleneck for software teams, and the rise of AI-generated code is making the problem worse. Reviewing large volumes of code slows devs down, and increases the chances of bugs slipping through. That's why we're building cubic: an AI code reviewer designed for complex codebases. Teams who adopt cubic merge merge PRs 48% faster. How? When you open a PR, cubic automatically reviews it. It learns from your codebase's patterns, points out bugs, suggesting fixes, and frees you from low-level review tasks.
Speclint
Score and fix user stories for AI coding agents.
Speclint is an open-source spec linter that scores user stories and requirements on a 0-100 scale across 5 dimensions: Measurable Outcomes (20pts), Testable Criteria (25pts), Constraints (20pts), No Vague Verbs (20pts), and Verification Steps (15pts). The scoring is fully deterministic — pure regex pattern matching, no LLM involved. AI is only used for the optional rewrite feature, which rewrites failing specs to pass all checks. Available as: CLI (npx speclint lint), GitHub Action, MCP Server (3 tools), and REST API. Free tier: 5 lints/day. Paid plans start at $9/mo for unlimited linting and rewrites. Open source (MIT). Built for teams shipping to AI coding agents — if your spec scores below 70, the agent will hallucinate. Fix the spec first.
Multitui
Create Custom Terminal Apps for macOS
Multitui is a macOS app factory designed to convert Terminal User Interface (TUI) programs into dedicated, native macOS applications. It allows users to wrap command-line tools like Claude Code, lazygit, htop, and vim into individual .app bundles with their own dock icons and window management. A standout feature is its integrated sandboxing, which leverages macOS's built-in sandbox-exec to provide fine-grained filesystem protection. This ensures that CLI tools or AI agents can run with restricted permissions, preventing them from deleting or modifying sensitive data unless explicitly allowed. Each app maintains its own shell history, snippets, and visual themes, providing a first-class macOS experience for power users.
Snippfy
Automates tedious dev work while keeping teams in control
Snippfy automates tedious dev work while keeping teams in control. It connects to task management tools like Jira & Trello, pulls tickets, writes high-quality code using AI, and creates PRs—so developers can focus on problem-solving, not busywork. Snippfy is an AI-powered automation tool for developers.
Agentset
Build AI apps that deliver reliable answers. No RAG expertise needed.
Agentset is an open-source RAG (Retrieval-Augmented Generation) infrastructure designed to help developers build production-grade AI chat and search applications. It handles the complex aspects of RAG, including document extraction, chunking, hybrid search, and agentic reasoning, ensuring that AI apps remain reliable even with large-scale data and high user traffic. The platform is model-agnostic, allowing users to choose their own LLMs, vector databases, and embedding models. It supports over 22 file formats and multimodal data like images and tables, providing cited answers out of the box for sectors like medical AI, legal tech, and enterprise search.