MCP servers (108)
Model Context Protocol servers expose tools, data and APIs to AI assistants through a single standard interface. This collection covers servers, clients and the libraries used to build them.
Collections
33 ranked collections built from the imported GitHub catalog. Rankings use recorded stars, not testing — and every project shows whether a commit has passed security review.
Model Context Protocol servers expose tools, data and APIs to AI assistants through a single standard interface. This collection covers servers, clients and the libraries used to build them.
Skills, plugins, commands and marketplaces that extend Claude Code. Expect packaged instructions, hooks and subagent workflows rather than standalone applications.
Agents that read, write and ship code — terminal agents, IDE agents and autonomous pull-request bots. Capability claims here are the repositories’ own; none of them are benchmarked by Techandria.
Frameworks and runtimes for building autonomous and multi-agent systems: planning loops, tool calling, memory and orchestration.
Conversational interfaces you can self-host or embed — chat UIs, assistant shells and the front ends people put in front of a model.
Prompt libraries, evaluation harnesses, playgrounds and versioning tools for teams that treat prompts as source code.
Retrieval-augmented generation stacks: chunking, embeddings, vector stores, rerankers and the glue that connects them to a model.
Runtimes, servers and desktop apps for running models on your own hardware — quantisation, inference engines and offline-first assistants.
Open-weight model families and the repositories that train, serve or fine-tune them.
Training and inference frameworks, model libraries and the numerical stacks underneath them.
Diffusion pipelines, video models, editors and the interfaces built on top of them.
Speech recognition, text to speech, voice cloning and audio understanding.
Component libraries and headless primitives for React applications — accessible building blocks rather than full templates.
Design systems and UI kits: tokens, themes, documentation and the components that ship with them.
Registries, blocks, themes and extensions built on shadcn/ui and Tailwind CSS.
Starters, plugins, adapters and libraries built specifically for the Next.js App Router and its ecosystem.
Animation engines, motion primitives, scroll effects and the CSS tooling around them.
Linters, audits, testing harnesses and accessible component primitives for WCAG work.
Component libraries, frameworks and tooling for Vue and Nuxt applications.
Headless browsers, crawlers and structured-extraction pipelines. Check each project’s terms and the target site’s rules before you run anything.
Visual builders, job runners and integration platforms for connecting services and scheduling work.
Command-line utilities, TUIs and shell tooling for people who live in a terminal.
Applications designed to run on your own infrastructure, with your own data.
Containers, orchestration, CI, observability and the platform layer underneath applications.
Databases, query layers, migration tools and analytics engines.
Authentication, secrets management, scanning and hardening. Listing here is discovery metadata — it is not an endorsement of a project’s own security.
Note apps, personal wikis, markdown editors and knowledge bases — including the AI layers people bolt onto them.
iOS, Android and cross-platform frameworks, component kits and developer tooling.
Native and hybrid desktop applications, plus the frameworks used to build them.
Python libraries, frameworks and applications across the AI and developer-tooling catalog.
TypeScript libraries, frameworks and applications across the AI and developer-tooling catalog.
Rust tools, runtimes and libraries in the catalog.
Go services, CLIs and libraries in the catalog.