Packages

Retry abstraction for Spring applications — provides declarative (@Retryable) and programmatic (RetryTemplate) retry with configurable backoff policies. Spring Retry features: @Retryable(retryFor=IOException.class, maxAttempts=3, backoff=@Backoff(delay=1000, multiplier=2)) annotation, @Recover fallback method, RetryTemplate with ExponentialBackOffPolicy, SimpleRetryPolicy, CircuitBreakerRetryPolicy, stateful retry for message-driven recovery, RetryContext for cross-attempt state, and @EnableRetry on configuration class. Used in Spring Batch, Spring AMQP, and Spring Kafka internally. Lighter weight than Resilience4j for pure retry use cases.

Spring framework for implementing finite state machines in Java/Kotlin. Spring State Machine models: States (IDLE, RUNNING, FAILED, COMPLETED), Events (START, COMPLETE, FAIL, RESET), Transitions (IDLE → RUNNING on START event), Guards (conditional transitions), Actions (execute code on state entry/exit/transition), and hierarchical/orthogonal state machines. Supports persistence (JPA, Redis, MongoDB) for distributed stateful workflows. Spring Boot auto-configuration for stateMachineFactory bean. Used for modeling agent workflow states, order processing pipelines, document approval workflows, and any process with discrete states and transitions.

Apple's modern Swift-native persistence framework, introduced in iOS 17 as the declared replacement for CoreData. SwiftData uses Swift macros (@Model, @Relationship, @Attribute) to define persistent model classes — no manual entity descriptions, no NSManagedObjectContext boilerplate. Integrates tightly with SwiftUI via @Query macro for reactive data fetching. Backed by CoreData under the hood but with a dramatically simpler Swift-native API.

Telegram Bot API — free REST API for building Telegram bots that can send messages, images, files, and interactive content to users, groups, and channels, enabling agents to deliver notifications, handle user interactions, and automate Telegram workflows.

The Odds API sports betting odds data REST API for developers, sportsbook comparison sites, and sports analytics platforms to access real-time and historical betting odds from 70+ sportsbooks across 40+ sports including NFL, NBA, MLB, NHL, soccer, tennis, and esports for odds comparison, arbitrage detection, and sports betting analytics. Enables AI agents to manage multi-sportsbook odds retrieval for betting comparison automation, handle live in-play odds for real-time sports betting automation, access historical odds for backtesting and analytics automation, retrieve arbitrage opportunity detection for value betting automation, manage sports market list for available event discovery automation, handle head-to-head and spread odds for betting market automation, access over/under and totals for advanced betting markets automation, retrieve player prop odds for individual performance betting automation, manage line movement tracking for sharp money detection automation, and integrate odds data with sports betting analytics platforms and comparison sites for end-to-end sports betting data automation.

Tinybird real-time analytics API platform — ingest events and publish low-latency SQL query results as HTTP API endpoints powered by ClickHouse under the hood.

Unsplash REST API — access to over 3 million high-resolution royalty-free photos for search, download, and collection management, enabling agents to automatically find and use professional photography in content creation workflows.

HTTP interaction recording and replay for Ruby tests — records real HTTP requests to cassette YAML files and replays them in subsequent test runs. VCR features: VCR.use_cassette('cassette_name') { } block wraps HTTP calls, record modes (:once, :new_episodes, :none, :all), WebMock/Faraday/Typhoeus adapter integration, cassette filtering for credentials (filter_sensitive_data), request matching customization (uri, body, headers), dynamic ERB cassettes, ignore hosts configuration, and before_record/before_playback hooks. Eliminates live HTTP calls in tests while keeping tests realistic with real response structure.

Kubernetes backup, restore, and migration tool maintained by VMware Tanzu. Velero backs up Kubernetes cluster resources (all objects from etcd) and persistent volume data to object storage (S3, GCS, Azure Blob), enabling disaster recovery, cluster migration, and namespace-level restore. Supports scheduled backups, pre/post-backup hooks, and selective resource filtering. The de-facto standard for Kubernetes backup.

High-performance, cost-efficient time series database and monitoring solution. VictoriaMetrics is a drop-in Prometheus replacement that offers 10x better compression, significantly lower resource usage, and horizontal scalability via VictoriaMetrics Cluster. Supports Prometheus remote write, PromQL, MetricsQL (extended query language), and native data ingestion formats. Often used as a long-term storage backend for Prometheus, replacing Thanos or Cortex. Powers large-scale monitoring at companies like Roblox, Cloudflare, and others.

HTTP session middleware for Express.js. Manages server-side sessions: creates a session identifier, stores it in a signed cookie, and provides req.session for storing user state server-side. The session store is pluggable — defaults to in-memory (not for production), with adapters for Redis (connect-redis), MongoDB (connect-mongo), PostgreSQL, and more. The standard approach for cookie-session-based auth in traditional Express web applications.

Geocoding library for Python — convert addresses to coordinates and back, calculate distances. geopy features: Nominatim geocoder (OpenStreetMap, free), GoogleV3 geocoder, ArcGIS, Bing, HERE, and 20+ other geocoders, geocode() for address→coordinates, reverse() for coordinates→address, GeocoderTimedOut handling, geopy.distance.geodesic() for great-circle distance calculation, geopy.distance.great_circle() for haversine, and User-Agent rate limiting compliance. Unified interface across 20+ geocoding providers — switch backends with one line change.

CLI tool that bulk port-forwards Kubernetes services to your local workstation, assigning unique loopback IPs per service and updating /etc/hosts so you can access cluster services by name as if they were running locally. Includes TUI, REST API, and MCP server integration.

A universal command-line client for the Model Context Protocol (MCP) that provides interactive shell, scripting, and AI agent integration with MCP servers over HTTP or stdio, with OAuth 2.1 support and secure credential storage.

All-in-one Docker-based sandbox environment for AI agents combining browser automation, shell execution, file operations, MCP servers, and VSCode Server in a single container with a unified file system.

AWS CDK lets you define cloud infrastructure as code in TypeScript, Python, Java, or Go, compiled to CloudFormation templates.

Blockchain infrastructure platform providing enhanced Ethereum/EVM node RPC access, NFT APIs, token APIs, and real-time event webhooks with best-in-class reliability and developer experience.

Google's open-source polyglot build system for large monorepos featuring hermetic builds, incremental compilation, and remote caching for reproducible artifacts.

NVIDIA conversation safety framework that uses the Colang language to define topical, fact-checking, and jailbreak-detection rails that control dialogue flow in LLM applications.

In-browser Node.js runtime built on WebAssembly. Runs a full Node.js environment directly in the browser tab without servers — including npm package installation, file system, and process execution. Powers bolt.new (AI-driven full-stack app builder), StackBlitz, and in-browser coding environments. Agent use: create isolated Node.js sandboxes for code execution without cloud infrastructure.