Curated AI & LLM tools.

Language for controlling large language models. Mix templating, control flow and generation in one expression.

Framework for programming — rather than prompting — language models. Compiles LLM pipelines into self-improving, reproducible programs.

Open-source embeddings database for AI apps. Run in-process, in a server, or in the cloud.

Production-ready LLM framework focused on RAG pipelines. Modular components for retrieval, generation and evaluation.

GitHub's official MCP server. Lets Claude (or any MCP client) browse repos, open issues, create PRs, and triage from inside the chat.

Open-source autonomous-agent framework with a polished GUI for designing, running and managing AI agents.

Stateful agents with long-term memory. OS-inspired hierarchical memory lets agents remember across sessions far beyond the context window.

OpenAI's framework for benchmarking LLMs and an open-source registry of evals. Industry-standard test harness.

Open-source Zapier alternative. AI-aware connectors, self-host or cloud, build automations in your browser.

AI browser automation framework in TypeScript. Combines deterministic code with natural-language actions — best of both worlds.

Access large language models from the command-line

Generate structured text and JSON from any LLM. Guarantees the output matches your schema or grammar.

Terminal-based AI dev tool focused on planning and executing large multi-file changes. Branching, version control built-in.

AI-native vector database. Combines vector search with structured filtering and GraphQL-style queries.

Open-source LLM engineering platform — tracing, prompt management, evaluations, datasets, playground.

Command-line productivity tool powered by AI. Describe what you want, sgpt suggests the right shell command.

Open-source AI coding agent built for the terminal. Beautiful TUI, model-agnostic, project-aware.

Pre-processing library for AI pipelines — extract text and tables from PDFs, Word, HTML, EML and 25+ other formats.

Config-driven LLM fine-tuning. YAML in, fine-tuned model out — abstracts away the boilerplate of every major training stack.

Type-safe agent framework from the Pydantic team. Structured outputs, dependency injection, model-agnostic.

Build real-time voice and video AI agents. Plug in any LLM + TTS + STT — LiveKit handles the WebRTC plumbing.

CLI and library for evaluating, testing, and red-teaming LLM apps. Side-by-side prompt comparisons in your CI.

AI-powered E2E testing in natural language. Write tests like "log in then click checkout" — the agent figures out the selectors.

Open-source LLM observability — traces, evaluation, datasets, retrieval debugging. OpenTelemetry-native.