Explore the directory

A curated, weekly link-checked list of real applications built through vibe coding, where a developer describes intent in natural language and an AI tool such as Claude Code, Cursor or Lovable generates and refines the code.

• Curated lists

A framework that has two AI models argue over a piece of code, with a third cross-validating, so a security or correctness finding survives adversarial attack instead of resting on one model's opinion.

• Code quality
• Security

A methodology and set of AI skills that turns a PRD (product requirements document) directly into code through a multi-agent system of specialized roles, checking completeness, correctness and coherence along the way.

• Agent skills

An on-device shell-command generator for macOS Tahoe that turns a plain-English description into the actual command, using Apple's built-in 3-billion-parameter model with few-shot examples drawn from a bank of 21,000 tldr command entries.

• CLI tools

A Swift and MLX implementation of the GLiNER2 information-extraction model, running named entity recognition, text classification, structured JSON extraction and relation extraction entirely on-device on Apple Silicon.

• Inference
• Libraries

A JSON Schema and C# object model for an exchange format that describes the metadata needed to generate data warehouse transformation logic, so different data-automation tools can share the same source-to-target and physical-model metadata.

• Data engineering

A command-line toolkit for security-testing Retrieval Augmented Generation pipelines, covering 27 techniques across six kill-chain phases mapped to MITRE ATLAS, from detecting a RAG pipeline to exfiltrating or poisoning its knowledge base.

• Security

A minimal runtime for running a recursive coding-agent loop over long tasks, where a short bash loop repeats agent turns until a stop file appears, and an outer meta-loop restarts a fresh generation when the current one stalls.

• Agent frameworks

An autonomous terminal agent that turns natural-language intent into verified system actions on Windows, Linux and Mac, with an --auto mode that executes, checks its own output and retries until a task completes.

• Coding agents

A Visual Studio Code extension that adds language support for Antlers, the templating language used to build Statamic websites, including syntax highlighting, semantic highlighting and completions based on a project's own blueprints and collections.

• Editors

A Python-first vector database with a Rust storage and search core, designed to move from an embedded local prototype to an HTTP service and then a lightweight sharded cluster without changing the client API.

• Databases

A browser-native inspector for neural network model files that opens a model and explores its computation graph, metadata, tensor shapes and weight statistics with no upload and no server.

• Models
• Debugging

A single-user, always-on personal AI assistant built on a Claude subscription that takes voice or text tasks across a TUI, CLI and self-hosted web app, runs scheduled jobs, and keeps long-term memory in plain Markdown.

• Agent frameworks

An Android app for running large language models entirely on-device, so a downloaded model can be used to chat fully offline with no API key, server, or data leaving the phone.

• Inference
• Mobile

A local MCP server that recommends the few right skills for a coding agent's current task out of a large skill collection, instead of loading the whole collection into context, and learns from whether each recommendation worked.

• MCP servers
• Agent skills

A command-line interface for Redmine project management that ships an agent skill and a built-in MCP server so an AI agent can list issues, view them and log time the same way a person would.

• MCP servers
• CLI tools

A personal AI agent written in pure Python, reachable from a terminal, a web dashboard, or Telegram, Discord and WhatsApp, that remembers what matters, learns new skills on demand, and runs tasks on a schedule.

• Agent frameworks

A user-owned, model-independent memory layer for coding agents that stores durable evidence, decisions and outcomes in PostgreSQL with pgvector and assembles compact context for the next task.

• Agent memory

A browser-based control panel for downloading and running large language models with vLLM or llama.cpp, rating each setting against the user's own hardware so a model launches without hours of trial and error.

• Inference

A Qt and C++ library that gives a desktop application cloud and local LLM providers, MCP clients and servers, and ACP agent hosting, with streaming delivered as ordinary Qt signals.

• Agent frameworks
• Libraries

A self-hosted, multi-agent personal assistant that reads a user's mail, calendar, contacts and documents through their own accounts, asks before making any change, and keeps a record of everything it read, decided and did.

• Agent frameworks

An MCP server that checks whether a component an AI agent is about to write already exists as a maintained library on GitHub, so the agent borrows, adapts or writes it on purpose instead of quietly rebuilding it.

• MCP servers
• Code quality

A toolkit that converts a dense Gemma-3-270m model into a mixture-of-experts model, then trains, inspects and serves it on CPU through llama.cpp.

• Models
• Inference

A local-first autonomous coding agent that plans, executes and verifies multi-step programming tasks on a user's own GPU through Ollama, optionally escalating to a larger cloud model for harder steps.

• Coding agents