Packages

A VS Code extension that runs an MCP server locally, exposing VS Code editing, file management, code intelligence (symbols, diagnostics, hover), and terminal execution to AI agents like Claude.

MCP server for Apple Voice Memos on macOS enabling AI agents to list, search, and access voice memo recordings from the local Voice Memos app. Allows AI agents to retrieve transcriptions and metadata from voice recordings stored locally on macOS for productivity and note-taking workflows.

Desktop application that extends Claude with persistent background automation, scheduled task execution, memory management, file watching, and worker agents. Runs as a menu bar/system tray app with a Node.js sidecar for scheduling and an MCP server for Claude integration. Supports pre-configured worker personalities (researcher, code reviewer, email assistant, etc.) and self-improving tasks that optimize based on execution history.

Improved child_process for Node.js with better developer experience. Execa wraps Node.js's child_process module with Promise-based API, proper error handling, piping, streaming, and cross-platform compatibility. Unlike zx (shell scripting focus), Execa is a pure Node.js API for spawning processes from code — no shell injection risk, better performance, and more control over process I/O.

MCP server wrapper around GPT Researcher that exposes deep research, quick search, report writing, and source retrieval as MCP tools. Supports STDIO, SSE, and Streamable HTTP transports for integration with Claude Desktop and other MCP clients.

MCP server integrating Google Gemini's search grounding capability — enabling AI agents to perform web searches using Google's search infrastructure via the Gemini API, with results grounded in real-time web content. Provides up-to-date information search with Google's index quality and Gemini's synthesis capabilities for research and information retrieval tasks.

Property-based testing library for Python that automatically generates hundreds of diverse test inputs — including edge cases humans miss — by describing the shape of valid inputs rather than writing specific examples, then shrinks failing cases to minimal reproducible examples.

The standard testing framework for Java and Kotlin on the JVM. JUnit 5 (Jupiter) provides annotations (@Test, @BeforeEach, @AfterEach, @Nested, @ParameterizedTest), assertions (assertEquals, assertThrows, assertAll), and an extension model for test customization. Replaced JUnit 4 as the de facto standard for JVM testing. Supports parameterized tests, dynamic tests, test templates, and parallel execution.

JavaScript testing framework with batteries included — test runner, assertion library, mocking, coverage, and snapshot testing in one package. Originally created by Meta (Facebook) for React. Jest uses a JSDOM environment by default, making React component testing straightforward. The most widely deployed JavaScript test framework with massive ecosystem compatibility.

Ecommerce-focused email and SMS marketing platform with deep Shopify/BigCommerce integration, behavioral segmentation, and APIs for event tracking, profile management, and campaign automation.

MCP Android Gateway server enabling AI agents to interact with Android devices — executing ADB commands, controlling Android apps via UI automation, capturing screenshots, installing and managing APKs, reading device logs, and integrating Android device control into agent-driven mobile development, testing, and automation workflows.

MCP server by prolific Chinese MCP developer shadowcz007 that enables AI agents to run local executables and programs. Provides a bridge between MCP-based agents and local system executables, allowing agents to invoke local tools, scripts, and applications via structured MCP tool calls rather than direct shell access.

Community MCP server for OpenTelemetry — enabling AI agents to query traces, spans, metrics, and logs from OpenTelemetry-compatible backends (Jaeger, Tempo, Prometheus, and OTLP-compatible stores).

Postman's programmatic API for managing collections, environments, workspaces, monitors, and mock servers — enabling automation of API testing, documentation, and collaboration workflows.

Python-native workflow orchestration platform that schedules, monitors, and recovers data pipelines and automation workflows, with a REST API for flow run management and observability.

Python library for encoding and decoding JSON Web Tokens (JWT). PyJWT handles creating signed JWTs (using HS256, RS256, ES256, etc.), verifying signatures, validating standard claims (exp, nbf, iss, aud), and decoding token payloads. Used in FastAPI auth middleware, OAuth2 token validation, and API key systems that use JWTs for agent authentication.

Type-safe Python AI agent framework built on Pydantic that enforces runtime validation of LLM outputs, provides structured result models, dependency injection for testability, and multi-provider LLM support.

MCP server providing reverse engineering and binary analysis capabilities to AI agents — enabling agents to disassemble binaries, analyze executable structures, extract strings, identify function signatures, and assist with malware analysis and security research workflows through structured MCP tool calls.

Community MCP server for Salesforce CRM enabling AI agents to query Salesforce objects via SOQL, manage records (leads, contacts, opportunities), and interact with Salesforce data in agent sales workflows.

Python async library implementing structured concurrency as a first-class principle. Trio's nursery API requires all spawned tasks to complete before their scope exits, preventing resource leaks and making cancellation predictable. An alternative to asyncio with a stricter, cleaner design. Compatible with asyncio via anyio bridge. Used where asyncio's complexity causes correctness issues.