Packages

JavaScript build tool and task runner, inspired by Rake (Ruby) and Make. Jake defines tasks in a Jakefile — plain JavaScript — with dependencies, namespacing, and file tasks. Less popular than Gulp/Grunt in modern projects but still used in established Node.js codebases. Provides a simple programmatic build system for Node.js projects without YAML configuration.

Provides access to Microsoft's Phi-4, a 14B-parameter small language model optimized for mathematical reasoning, STEM tasks, and instruction following, via the Azure AI Inference API in Azure AI Studio or Azure AI Foundry.

Enterprise identity platform providing SSO, MFA, and lifecycle management for users and applications via REST API and OAuth2/OIDC.

Python SSHv2 library for securely connecting to remote machines to execute commands and transfer files programmatically.

Authentication middleware for Node.js with a strategy plugin system supporting 500+ authentication strategies (Local, OAuth, Google, GitHub, Facebook, SAML, etc.). Passport handles authentication flow, session management, and user serialization/deserialization. Works with Express and Connect-based frameworks. The standard authentication library in the Node.js ecosystem.

Stitch (Talend) cloud data integration REST API for ELT pipelines from 130+ sources to cloud data warehouses. Enables AI agents to manage data source connection and replication configuration for automated data pipeline setup, handle extraction job scheduling and monitoring for data pipeline operations, access replication progress and row count reporting for pipeline visibility, retrieve connection health and error monitoring for pipeline reliability management, manage destination configuration (Snowflake, Redshift, BigQuery, Databricks) for data warehouse automation, handle incremental replication key configuration for efficient data extraction, access schema discovery and field selection for selective data extraction, retrieve extraction logs and debugging data for pipeline troubleshooting automation, manage API source configuration for custom data source integration, and integrate Stitch with data catalog, dbt, and analytics platforms for modern data stack automation.

Web3 development platform that enables AI agents to deploy smart contracts, mint NFTs, authenticate wallets, and execute gasless blockchain transactions via REST API and TypeScript/Python SDKs.

Production-grade model serving framework for PyTorch models developed by AWS and Meta/Facebook. TorchServe provides REST and gRPC APIs for serving PyTorch models with features including multi-model serving, model versioning, A/B testing, batch inference, metrics, and dynamic model loading/unloading without server restarts. Acts as the reference serving solution for the PyTorch ecosystem — used when you have PyTorch models and need a production inference server without building one from scratch.

Open-source vector database with built-in ML vectorization modules (text2vec, multimodal) accessible via REST, GraphQL, and gRPC, available as self-hosted or managed Weaviate Cloud.

WooCommerce's REST API exposes store management capabilities for a WordPress-based e-commerce site, including products, orders, customers, coupons, reports, and settings. The API is embedded in the WordPress site itself (wp-json/wc/v3).

Advanced caching library for Python that prevents cache stampede — provides a region-based caching system with dogpile lock to prevent multiple threads from regenerating the same expired cache entry simultaneously. dogpile.cache features: CacheRegion with make_region() factory, @cache_region.cache_on_arguments() decorator, mutex-based stampede prevention (one thread regenerates while others serve stale data), Redis/Memcached/file/memory backends, key mangling, expiration with grace period (serve stale while refreshing), async cache refresh support, invalidation, and integration with SQLAlchemy ORM. From the SQLAlchemy project.

Serverless AI model inference platform offering sub-second image, video, and audio generation via REST API, with support for Flux, SDXL, Wan, and hundreds of open-source models.

Cloud-native fine-grained authorization service for applications and APIs. Aserto provides a hosted Open Policy Agent (OPA) service with user/group management, role-based access control (RBAC), and relationship-based access control (ReBAC / Google Zanzibar model). Decision logs, policy versioning, and middleware SDKs for Express, FastAPI, Rails, and more. Agents call Aserto's authorization API to answer 'can user X perform action Y on resource Z?' without building custom authorization logic.

Fully managed, globally distributed NoSQL and relational database service from Microsoft Azure. Supports multiple APIs (NoSQL/Core, MongoDB, Cassandra, Gremlin, Table) over the same underlying engine. Offers single-digit millisecond reads, configurable consistency levels (5 levels from eventual to strong), and turnkey multi-region writes with automatic failover.

Accelerates connections from Cloudflare Workers to remote PostgreSQL databases by maintaining persistent regional connection pools and caching query results at the edge.

Couchbase REST API enables agents to perform distributed NoSQL document operations, N1QL/SQL++ queries, full-text search, and mobile data sync against a high-performance multi-service cluster.

Finance Trading AI Agents MCP server enabling AI agents to access financial market data and trading tools — fetching stock prices and market data, analyzing financial instruments, and integrating market data into agent-driven investment research and trading analysis workflows.

Hugging Face Inference API — run inference on 200,000+ open-source models (LLMs, NLP, CV, audio) via a unified REST API without managing GPU infrastructure.

HuggingFace Datasets library — fast, efficient loading and processing of ML datasets. Datasets features: load_dataset() for 100,000+ public datasets from Hub, local file loading (CSV, JSON, Parquet, text), Dataset.map() for preprocessing, Dataset.filter(), streaming mode (streaming=True for infinite datasets), Arrow-backed memory mapping (no RAM limits), multi-GPU/CPU parallelism, push_to_hub() for sharing, DatasetDict for train/validation/test splits, dataset caching, and Sequence/ClassLabel/Value feature types. Core library for agent LLM training data, evaluation benchmarks, and RAG document ingestion.

Systems programming language designed as a superset of Python, created by Modular (Chris Lattner). Mojo extends Python with manual memory management, SIMD vectorization, GPU programming, and MLIR compiler infrastructure for ML workloads. Goal: Python syntax with C++/CUDA performance for ML code without leaving the Python ecosystem. Mojo compiles to native code and can call Python libraries. Powers Modular's MAX engine for ML inference acceleration.