I want to look up a package

Agentor framework includes LiteMCP, a native ASGI MCP server that integrates directly with FastAPI using standard patterns. Supports decorator-based tool definitions with built-in CORS and authentication controls.

Performance-oriented multi-agent framework with built-in memory, knowledge, and multi-modal support.

Async-first HTTP client and server framework.

Mature, extensible platform to author, schedule, and monitor workflows as DAGs with a rich UI and huge operator ecosystem.

Declarative statistical visualization library based on the Vega-Lite grammar — concise, expressive grammar of graphics for Python.

Approximate Nearest Neighbor Oh-Yeah — lightweight C++/Python library for fast, memory-efficient ANN search with static indexes (good for embedding similarity).

Official Anthropic Python client for the Claude API — messages, streaming, tool use, prompt caching, and AWS Bedrock / Vertex AI integration.

Async networking and concurrency library that runs on top of asyncio or Trio — provides Trio-style structured concurrency primitives with broad compatibility.

Python's built-in command-line parsing library (standard library); supports options, flags, subcommands, type conversion, and help text.

Open-source LLM observability and evaluation platform from Arize — built on OpenTelemetry, runs locally or self-hosted with rich tracing UI.

Built-in Python library for writing concurrent code using async/await syntax, providing the default event loop and task scheduling system.

Class boilerplate generator that produces well-behaved data classes with validators, converters, and slots — predates and complements dataclasses.

Microsoft framework for multi-agent conversations, workflows, and tool use.

Python library for pulling data out of HTML and XML files; pairs well with requests, lxml, or html.parser.

Open-source Python framework for packaging, serving, and deploying ML/LLM models as production APIs — model-agnostic with built-in adaptive batching.

The uncompromising Python code formatter — produces consistent, deterministic output with minimal config.

Interactive visualization library for modern web browsers — handles large/streaming datasets and produces standalone HTML or app dashboards.

Open-source Python library for building AI agents that drive browsers — give it a goal in natural language and an LLM figures out the clicks, scrolls, and form fills via Playwright.

Lightweight, fast text chunking library for RAG pipelines with multiple strategies (token, sentence, semantic, neural) and 32+ integrations.

Lightweight open-source embedding database built for local, serverless, or cloud-backed vector search — ideal for prototyping RAG apps.

Composable Python package for creating beautiful command-line interfaces with minimal code (options, prompts, subcommands).

LLM-friendly web crawler and scraper that returns clean markdown/JSON optimized for RAG and LLM ingestion — async-first with browser automation.

Framework for multi-agent "crews" with roles and tools for collaborative tasks.

Modern data orchestrator with software-defined assets, type-checked I/O, and rich UI for data platforms.

A Python framework for building reactive web-apps. Developed by Plotly.

Parallel and distributed computing for Python — scale pandas/numpy beyond memory across cores or clusters.

Built-in Python module for lightweight data structures — provides boilerplate-free class creation with type hints and defaults.

Synthetic text dataset generator for LLM projects supporting classification, instruction, MCQ, and preference datasets via multiple LLM providers.

Pytest-style framework for evaluating LLM outputs with built-in metrics — hallucination, answer relevancy, faithfulness, G-Eval, and more.

Batteries-included web framework with ORM, auth, and admin.

Open-source toolkit by IBM Research for parsing diverse document formats (PDF, DOCX, PPTX, HTML, XLSX, images, audio) into a unified structured representation ideal for GenAI and RAG workflows. Hosted by LF AI & Data Foundation.

Stanford framework for programming LLMs declaratively with signatures, modules, and prompt optimizers.

Meta's library for efficient similarity search and clustering of dense vectors — the indexing primitive behind many vector DBs.

Popular library for generating fake but realistic data such as names, addresses, emails, dates, text, and more for testing and prototyping.

High-performance async APIs with automatic OpenAPI docs.

Lightweight ONNX-based embedding library by Qdrant — runs on CPU with no PyTorch dependency, designed for fast batch inference in production RAG pipelines.

The standard Python framework for building MCP servers and clients.

Google's library for automatically generating CLIs from any Python object — turns functions, classes, or modules into command-line tools.

Embedding and reranking library from BAAI — home of the BGE family of state-of-the-art retrieval models with multilingual, multimodal, and long-context variants.

Wraps pycodestyle, pyflakes, and McCabe complexity into a single command-line linter with a plugin system.

Minimal, flexible microframework with a massive ecosystem.

Classic fuzzy string matching library based on Levenshtein distance; largely superseded by RapidFuzz and archived in 2024.

Open-source testing & evaluation platform for ML/LLM — test suites, bias/safety checks, and regression testing with a web UI.

Lightweight open-source flexible and modular framework for developing and deploying AI agents. While optimized for Gemini and the Google ecosystem, ADK is model-agnostic, deployment-agnostic, and is built for compatibility with other frameworks

Build shareable ML/AI web apps and demos—fast.

Temporal knowledge-graph memory for AI agents — extracts entities and relationships from conversations into a Neo4j-backed graph that tracks how facts evolve over time.

Modern, extensible Python project manager from PyPA — environments, build, publish, and version management with plugin system.

End-to-end pipelines for RAG, search, and evaluation (v2 haystack-ai).

Requests-like client with sync + async and HTTP/2.

Property-based testing library that generates examples to find edge cases your example-based tests miss.

UK AI Safety Institute's framework for large-scale LLM evaluations — solvers, scorers, and rich logging for benchmarks and safety evals.

Structured outputs from LLMs using Pydantic — patches OpenAI/Anthropic/Cohere/etc. clients to return validated typed objects with retries.

Sorts and formats Python imports automatically by section and alphabetically — pairs naturally with formatters.

Embedded serverless vector database in Rust, built on the Lance columnar format — local-first with cloud sync.

Modular framework for building LLM apps and agent workflows(Includes LangGraph).

Text splitting utilities from the LangChain ecosystem supporting character, token, recursive, semantic, and code-aware chunking strategies.

Open-source observability & evaluation platform for LLM/agent applications — tracing, metrics, datasets, prompt versioning, UI, self-hostable.

Graph-based orchestration framework for building stateful, controllable LLM agents as nodes and edges.

Unified observability & evaluation platform for LLM/agent applications — tracing, dashboards, alerts, feedback & evaluation.

Stateful agent framework formerly known as MemGPT — gives agents long-term memory through self-managed context windows that page in and out of an SQL backend.

Unified Python SDK and proxy server that calls 100+ LLM providers using the OpenAI API format — drop-in router with cost tracking, fallbacks, and load balancing.

Type-first async framework with built-in OpenAPI.

Data framework focused on RAG—indexes, readers, retrievers, and workflows.

Text splitting and node parsing components from LlamaIndex for chunking documents into nodes optimized for indexing and retrieval in RAG workflows.

Observability platform from the Pydantic team built on OpenTelemetry, with native Python/Pydantic integration.

Python's built-in logging library for structured logging.

Delightful logging for Python with zero-config colorful output.

PDF/image to clean Markdown converter using deep-learning models — strong on tables, equations, and complex academic layouts.

Microsoft utility for converting Office docs, PDFs, images, audio, and more to Markdown — designed for LLM-friendly text extraction.

Object serialization and validation library for converting complex datatypes to and from native Python types and JSON.

Foundational Python plotting library providing low-level control over static, animated, and interactive visualizations.

Official Python SDK from the Model Context Protocol team for building MCP servers and clients — provides both high-level (FastMCP) and low-level primitives.

Self-improving memory layer for AI agents — stores user preferences, conversation history, and learned facts with automatic deduplication and recency scoring.

Open-source, high-performance vector database and indexing engine designed for billion-scale vector similarity search and analytics.

Fast and extensible fake data generator focused on performance, rich providers, and strict typing with excellent editor support.

Distillation framework that turns any sentence-transformer into a static word embedding model — 500x smaller and 500x faster than the teacher with minimal quality loss.

High-performance schema validation and serialization library for JSON, MessagePack, YAML, and TOML — written in C.

The original and canonical static type checker for Python — the PEP 484 reference implementation that defined the modern Python type-hint ecosystem.

Drop-in replacement for Requests with HTTP/2, HTTP/3, async, WebSocket, and SSE support.

Flexible test automation tool configured with Python scripts (noxfile.py) instead of INI files.

Foundational n-dimensional array library and the bedrock of scientific Python.

Official Python client for the Ollama local LLM runtime — pull, run, and chat with open-weight models on your machine.

Official OpenAI Python client for the OpenAI API — chat completions, embeddings, images, audio, and the new Responses + Assistants APIs.

BentoML-built CLI and Python framework for self-hosting open-weight LLMs as OpenAI-compatible API endpoints with one command.

Open-source OpenTelemetry-based observability for LLMs — vendor-neutral SDK that emits standard OTel traces consumable by any backend.

Open-source LLM evaluation and observability platform from Comet — tracing, evaluation datasets, prompt playground with self-hosted and cloud options.

Multilingual OCR toolkit with text detection, recognition, layout analysis, and table extraction — built on Baidu's PaddlePaddle framework.

Powerful data structures and data analysis tools for Python.

Flexible framework for dashboards and complex data apps (HoloViz ecosystem).

PostgreSQL extension that adds a native vector column type and indexing for similarity search in a relational database (ANN search inside Postgres).

Fully managed vector database for high-performance similarity search, optimized for production scale and hybrid search.

The official Python package installer — bundled with Python and the foundation for installing libraries from PyPI.

Installs and runs Python CLI applications in isolated environments — like pip for end-user tools rather than libraries.

Microsoft's modern browser automation library for Chromium, Firefox, and WebKit — async-first, with auto-waiting and powerful selectors.

Interactive, browser-based plotting library — produces publication-quality charts with hover, zoom, and 3D, often used with Dash for dashboards.

Batteries-included Python packaging & dependency manager with lockfiles, virtualenv management, building, and publishing.

Lightning-fast DataFrame library with lazy evaluation and multi-threading.

Pythonic workflow orchestration with lightweight flows/tasks, great developer ergonomics, and optional cloud UI.

Programmable Regular Expressions — object-based API for constructing, combining, and managing regex patterns in Python.

Data validation and settings management using Python type hints — creates schemas, validates data, and serializes models.

Type-safe agent framework from the Pydantic team with structured outputs and validation-first design.

Static code analyzer for Python that checks for errors, enforces coding standards (PEP8+), and suggests refactorings.

Fast, feature-rich PDF toolkit for text/HTML extraction, images, metadata, and page rendering with coordinates.

Text parsing and grammar framework that lets you define grammars in Python code — a readable alternative to raw regex for structured text.

Meta's high-performance Python type checker written in Rust — open-sourced in 2025 as the successor to Pyre, designed for monorepo-scale codebases.

Microsoft's fast static type checker for Python — the engine behind Pylance in VS Code, written in TypeScript with strong inference and IDE-first diagnostics.

Python API for Apache Spark — distributed dataframes and SQL for large-scale data processing.

Modern testing framework with powerful fixtures, plugins, and concise test syntax.

Fast C extension providing Levenshtein distance and related string similarity metrics; now maintained by the RapidFuzz project.

Open-source vector search engine with filtering, payload storage, and rich Python/REST/gRPC clients — Rust-built for production scale.

Metrics and tooling for evaluating RAG systems — faithfulness, answer relevance, context precision/recall, and dataset utilities.

High-performance fuzzy string matching library with C++ speed and a clean Python API; modern replacement for fuzzywuzzy.

Built-in Python module for regular expressions — provides search, match, replace, and pattern-based text processing.

Library for building user interfaces in Python without JavaScript — component-based and reactive, inspired by ReactJS. Supports multiple backends including Flask, FastAPI, Sanic, Tornado, Django, and Jupyter.

Human-friendly HTTP client.

Extremely fast Python linter and formatter written in Rust — consolidates rules from Flake8, isort, pyupgrade, and more.

Async-first Python web framework focused on speed via uvloop.

High-level, asynchronous web crawling & scraping framework with spiders, pipelines, middlewares, and selectors.

Synthetic Data Vault — generates statistically realistic synthetic datasets using probabilistic and deep learning models while preserving privacy.

High-level statistical data visualization library built on top of Matplotlib with attractive default styles.

The de facto standard for browser automation — drives any major browser via WebDriver protocol, mature ecosystem with grid for parallel execution.

De-facto standard Python library for computing sentence, paragraph, and image embeddings — wraps Hugging Face transformer models with a simple `encode()` API.

Production-grade browser-automation platform that uses vision LLMs to drive workflows like form filling, invoice processing, and data extraction — runs as a self-hosted or cloud server, not just a Python lib.

Python port of Browserbase's Stagehand — write deterministic Playwright code and reach for AI primitives like `act()`, `extract()`, and `observe()` only where you want LLM help.

Lightweight ASGI toolkit powering many async frameworks.

Turn Python scripts into interactive data web apps in minutes.

Structured logging library for Python that wraps stdlib logging with key-value context and pluggable processors.

Durable execution platform — write long-running workflows in Python with automatic retries, state persistence, and exactly-once semantics.

Unified Python library implementing 30+ string distance and similarity algorithms under one consistent API.

Mature non-blocking web framework built for long-lived connections and websockets.

Generic virtualenv management and test command-line tool that runs tests across multiple Python versions and environments.

Modern async concurrency library focused on structured concurrency, making async code easier to reason about and safer to write.

Astral's new ultra-fast Rust-based Python type checker from the makers of ruff and uv — pre-1.0 but already widely adopted for its speed.

Type-hint-driven CLI framework from the FastAPI author — automatic help, completion, and validation from Python type annotations.

Built-in Python testing framework inspired by Java's JUnit — part of the standard library.

Toolkit for loading and partitioning documents (PDFs, HTML, PPTX, images) into clean text/chunks with metadata for downstream NLP/RAG.

Low-level, battle-tested HTTP client powering many libraries.

Ultra-fast Python package & project manager (Rust) that replaces pip/virtualenv/pip-tools/pipx with a single, blazing-speed CLI.

High-performance drop-in replacement for Python's default asyncio event loop, built on libuv for faster asynchronous I/O.

High-throughput, memory-efficient LLM inference and serving engine with PagedAttention and continuous batching — built for production GPU workloads.

Open-source vector database and semantic search engine with hybrid queries, schema, and generative search.

Hosted long-term memory service for AI agents built on Graphiti — stores conversation history as a temporal knowledge graph and retrieves the most relevant context per turn.