Architecture on AI Knowledge Base

Emerging Architectures for LLM Applications

Mon, 01 Jan 0001 00:00:00 +0000

Reference architecture for the LLM application stack, covering data pipelines, embedding models, vector databases, orchestration frameworks, and operational tooling.

Introduction to Model Context Protocol

Mon, 01 Jan 0001 00:00:00 +0000

The only structured course on MCP available anywhere — Anthropic Academy’s introduction covers MCP architecture, the client-server model, tools/resources/prompts primitives, and building your first MCP server. This is the video/hands-on complement to the MCP documentation in this knowledge base. The MCP Fundamentals learning path provides the documentation reference; this course provides the guided walkthrough with exercises.

Symphony Service Specification

Mon, 01 Jan 0001 00:00:00 +0000

Complete technical specification for implementing Symphony: six architectural layers, orchestration state machine, workspace management, agent runner protocol, polling and scheduling, retry semantics, and observability.

MCP: Advanced Topics

Mon, 01 Jan 0001 00:00:00 +0000

Advanced MCP patterns including complex tool implementations, resource management, transport configuration, authentication, and production deployment. This is the second course in Anthropic’s unique MCP sequence — no other provider offers this depth of MCP training. Pairs with the Building MCP Servers learning path for a complete MCP education: the course provides guided exercises, the path provides documentation deep-dives.

MCP Fundamentals

Mon, 01 Jan 0001 00:00:00 +0000

Go from zero to a solid understanding of the Model Context Protocol. This path covers the core architecture, primitives (tools, resources, prompts), transport mechanisms, and how clients and servers interact.

By the end you’ll understand the MCP mental model well enough to evaluate MCP servers, build simple integrations, and read the specification confidently.

Orchestration State Machine

Mon, 01 Jan 0001 00:00:00 +0000

Symphony’s state machine for issue lifecycle: polling, dispatch, active run tracking, retry queuing with exponential backoff, and reconciliation of running issues against tracker state changes.

The Agentic Protocol Stack

Mon, 01 Jan 0001 00:00:00 +0000

Understand the three open protocols forming the agentic AI communication stack:

MCP (agent-to-tool) — how agents use external tools and data
A2A (agent-to-agent) — how agents collaborate across platforms
AG-UI (agent-to-user) — how agents communicate with frontends

All three are converging under the Linux Foundation’s Agentic AI Foundation (AAIF). This path teaches you the architecture and practical usage of each layer.

Deployment to Production

Mon, 01 Jan 0001 00:00:00 +0000

Take AI applications from prototype to production. This path covers the full deployment journey across providers: production checklists, rate limit management, batch processing for cost reduction, and agent-specific deployment patterns.

The key cross-provider insight: every provider has rate limits, batch APIs, and deployment recommendations, but the specifics differ meaningfully. Learning both OpenAI and Anthropic patterns helps you design resilient systems, and understanding framework deployment (LangGraph, CrewAI) is essential for agent-based applications.

Understanding AI Agents & Protocols

Mon, 01 Jan 0001 00:00:00 +0000

A non-technical guide to AI agents and the emerging protocol standards for product leaders and executives. Learn what agents are, why they matter for your product, and how the industry is standardizing agent communication through protocols like MCP, A2A, and AG-UI.

By the end of this path, you’ll understand the difference between chatbots and agents, how agents connect to external tools and data, and what enterprise agent deployment looks like. You’ll be equipped to evaluate whether your product needs agent capabilities and what infrastructure choices your team should be making.

AI Architecture for Technical Leaders

Mon, 01 Jan 0001 00:00:00 +0000

A technically rigorous guide to AI system architecture for product managers and executives who can evaluate architecture diagrams. This path walks through the major architectural decisions in modern AI systems — from the LLM application stack to agent frameworks, communication protocols, and multi-model strategies.

By the end of this path, you’ll be able to: evaluate your engineering team’s architecture proposals with informed questions, understand the protocol landscape (MCP, AG-UI) for platform integration decisions, and assess whether your system architecture is ready for multi-model, multi-agent production workloads.

A2A JSON-RPC

Mon, 01 Jan 0001 00:00:00 +0000

Communicate with an assistant using the Agent-to-Agent (A2A) Protocol over JSON-RPC 2.0.

A2A vs MCP

Mon, 01 Jan 0001 00:00:00 +0000

Understanding how A2A complements the Model Context Protocol

A2A vs MCP

Mon, 01 Jan 0001 00:00:00 +0000

Detailed comparison with Model Context Protocol

Access the W&B MCP Server

Mon, 01 Jan 0001 00:00:00 +0000

Connect your IDE or LLM application to W&B’s Model Context Protocol (MCP) server to provide your agent with access to your W&B workspace, data, and W&B’s documentation.

ADK with Agent2Agent (A2A) Protocol

Mon, 01 Jan 0001 00:00:00 +0000

Agent Client Protocol (ACP)

Mon, 01 Jan 0001 00:00:00 +0000

Expose Deep Agents over the Agent Client Protocol (ACP) to integrate with code editors and IDEs.

Agent Client Protocol (ACP)

Mon, 01 Jan 0001 00:00:00 +0000

Expose Deep Agents over the Agent Client Protocol (ACP) to integrate with code editors and IDEs.

Agent Framework Overview

Mon, 01 Jan 0001 00:00:00 +0000

Introduction to Microsoft Agent Framework: combining AutoGen’s agent abstractions with Semantic Kernel’s enterprise features for production-grade multi-agent orchestration.

Agents

Mon, 01 Jan 0001 00:00:00 +0000

Learn about agents in the Agent User Interaction Protocol

Agents Overview

Mon, 01 Jan 0001 00:00:00 +0000

Core agent abstraction in the Microsoft Agent Framework: agent types, providers, skills, and conversation management.

AI Mind Tool

Mon, 01 Jan 0001 00:00:00 +0000

The ‘AIMindTool’ is designed to query data sources in natural language.

Antitrust Policy

Mon, 01 Jan 0001 00:00:00 +0000

MCP Project Antitrust Policy for participants and contributors

API deployment checklist

Mon, 01 Jan 0001 00:00:00 +0000

Review commonly underused Responses API design choices that improve deployment quality, speed, cost, and reliability.

Architecture

Mon, 01 Jan 0001 00:00:00 +0000

Architecture

Mon, 01 Jan 0001 00:00:00 +0000

Chroma is designed with a modular architecture that prioritizes performance and ease of use. It scales seamlessly from local development to large-scale production, while exposing a consistent API across all deployment modes.

Architecture

Mon, 01 Jan 0001 00:00:00 +0000

Learn how Pinecone’s architecture enables fast, relevant vector search at any scale.

Architecture overview

Mon, 01 Jan 0001 00:00:00 +0000

The most important mental model here is the three-layer architecture: Host (the application), Client (the protocol handler), and Server (the capability provider). A common mistake is conflating the host and the client – the host is the user-facing application (like Claude Desktop), while the client is an internal protocol component that maintains a 1:1 connection with a single server. Understanding this separation is essential before reading the specification or building anything.

Architecture Overview

Mon, 01 Jan 0001 00:00:00 +0000

How Chroma is structured across local, single-node, and distributed deployments.

Authorization

Mon, 01 Jan 0001 00:00:00 +0000

The MCP authorization specification defines the OAuth 2.0-based flow that remote MCP servers use to authenticate clients, and it is the formal basis for everything the higher-level registry authentication guide simplifies. Focus on the required OAuth grant types, token scoping rules, and the metadata discovery mechanism that clients use to find a server’s authorization endpoint. A key gotcha is that the spec mandates specific OAuth metadata fields that many generic OAuth libraries do not populate by default, so you may need custom configuration even when using well-known auth frameworks. This is dense specification text best read alongside a working implementation for reference.

Authorization Extensions

Mon, 01 Jan 0001 00:00:00 +0000

Supplementary authorization mechanisms for the Model Context Protocol

Awesome A2A

Mon, 01 Jan 0001 00:00:00 +0000

Curated list of A2A resources

Brave Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘BraveSearchTool’ is designed to search the internet using the Brave Search API.

Build an MCP App

Mon, 01 Jan 0001 00:00:00 +0000

Getting started guide for building interactive UI applications with MCP Apps

Build an MCP client

Mon, 01 Jan 0001 00:00:00 +0000

Get started building your own client that can integrate with all MCP servers.

Build an MCP server

Mon, 01 Jan 0001 00:00:00 +0000

Get started building your own server to use in Claude for Desktop and other clients.

Build with Agent Skills

Mon, 01 Jan 0001 00:00:00 +0000

Use agent skills to guide AI coding assistants through MCP server design and implementation

Callback patterns

Mon, 01 Jan 0001 00:00:00 +0000

Cancellation

Mon, 01 Jan 0001 00:00:00 +0000

Capabilities

Mon, 01 Jan 0001 00:00:00 +0000

Dynamic capability discovery for agents in the AG-UI protocol

ChatKit widgets

Mon, 01 Jan 0001 00:00:00 +0000

Learn how to design widgets in your chat experience with ChatKit.

Code Docs RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘CodeDocsSearchTool’ is a powerful RAG (Retrieval-Augmented Generation) tool designed for semantic searches within code documentation.

Code Interpreter

Mon, 01 Jan 0001 00:00:00 +0000

The ‘CodeInterpreterTool’ is a powerful tool designed for executing Python 3 code within a secure, isolated environment.

Community

Mon, 01 Jan 0001 00:00:00 +0000

Join discussions and contribute to protocol development

Completion

Mon, 01 Jan 0001 00:00:00 +0000

Component architecture

Mon, 01 Jan 0001 00:00:00 +0000

Component architecture

Mon, 01 Jan 0001 00:00:00 +0000

Connect Claude Code to tools via MCP

Mon, 01 Jan 0001 00:00:00 +0000

Learn how to connect Claude Code to your tools with the Model Context Protocol.

Connect to local MCP servers

Mon, 01 Jan 0001 00:00:00 +0000

Learn how to extend Claude Desktop with local MCP servers to enable file system access and other powerful integrations

Connect to remote MCP Servers

Mon, 01 Jan 0001 00:00:00 +0000

Learn how to connect Claude to remote MCP servers and extend its capabilities with internet-hosted tools and data sources

Contributing

Mon, 01 Jan 0001 00:00:00 +0000

How to participate in Agent User Interaction Protocol development

Contributing to MCP

Mon, 01 Jan 0001 00:00:00 +0000

How to contribute to the Model Context Protocol project

Contributor Communication

Mon, 01 Jan 0001 00:00:00 +0000

Communication strategy and framework for the Model Context Protocol community

Contributor Ladder

Mon, 01 Jan 0001 00:00:00 +0000

Roles, responsibilities, and advancement criteria for MCP contributors, from first contribution to Core Maintainer

Core architecture

Mon, 01 Jan 0001 00:00:00 +0000

Understand how AG-UI connects front-end applications to AI agents

Crafting Effective Agents

Mon, 01 Jan 0001 00:00:00 +0000

Learn best practices for designing powerful, specialized AI agents that collaborate effectively to solve complex problems.

CSV RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘CSVSearchTool’ is a powerful RAG (Retrieval-Augmented Generation) tool designed for semantic searches within a CSV file’s content.

DALL-E Tool

Mon, 01 Jan 0001 00:00:00 +0000

The ‘DallETool’ is a powerful tool designed for generating images from textual descriptions.

Debugging

Mon, 01 Jan 0001 00:00:00 +0000

A comprehensive guide to debugging Agent User Interaction Protocol (AG-UI) integrations

Debugging

Mon, 01 Jan 0001 00:00:00 +0000

A comprehensive guide to debugging Model Context Protocol (MCP) integrations

Design Principles

Mon, 01 Jan 0001 00:00:00 +0000

The core design principles that guide the development of the Model Context Protocol.

Directory RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘DirectorySearchTool’ is a powerful RAG (Retrieval-Augmented Generation) tool designed for semantic searches within a directory’s content.

Directory Read

Mon, 01 Jan 0001 00:00:00 +0000

The ‘DirectoryReadTool’ is a powerful utility designed to provide a comprehensive listing of directory contents.

Distributed Architecture

Mon, 01 Jan 0001 00:00:00 +0000

How Chroma scales out with independent services, object storage, SSD caches, and a shared system database.

DOCX RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘DOCXSearchTool’ is a RAG tool designed for semantic searching within DOCX documents.

Elicitation

Mon, 01 Jan 0001 00:00:00 +0000

Enterprise-Managed Authorization

Mon, 01 Jan 0001 00:00:00 +0000

Centralized access control for MCP in enterprise environments via identity providers

Events

Mon, 01 Jan 0001 00:00:00 +0000

Understanding events in the Agent User Interaction Protocol

Events

Mon, 01 Jan 0001 00:00:00 +0000

Documentation for the events used in the Agent User Interaction Protocol SDK

Events

Mon, 01 Jan 0001 00:00:00 +0000

Documentation for the events used in the Agent User Interaction Protocol Python SDK

EXA Search Web Loader

Mon, 01 Jan 0001 00:00:00 +0000

The ‘EXASearchTool’ is designed to perform a semantic search for a specified query from a text’s content across the internet.

Example Clients

Mon, 01 Jan 0001 00:00:00 +0000

A list of applications that support MCP integrations

Example Servers

Mon, 01 Jan 0001 00:00:00 +0000

A list of example servers and implementations

Extension Support Matrix

Mon, 01 Jan 0001 00:00:00 +0000

Which MCP clients implement which official extensions

Extensions Overview

Mon, 01 Jan 0001 00:00:00 +0000

Optional extensions to the Model Context Protocol

File Read

Mon, 01 Jan 0001 00:00:00 +0000

The ‘FileReadTool’ is designed to read files from the local file system.

File Write

Mon, 01 Jan 0001 00:00:00 +0000

The ‘FileWriterTool’ is designed to write content to files.

Firecrawl Crawl Website

Mon, 01 Jan 0001 00:00:00 +0000

The ‘FirecrawlCrawlWebsiteTool’ is designed to crawl and convert websites into clean markdown or structured data.

Firecrawl Scrape Website

Mon, 01 Jan 0001 00:00:00 +0000

The ‘FirecrawlScrapeWebsiteTool’ is designed to scrape websites and convert them into clean markdown or structured data.

Frequently Asked Questions

Mon, 01 Jan 0001 00:00:00 +0000

This FAQ addresses the practical edge cases that the other registry guides do not cover in depth, including questions about naming conflicts, transfer of ownership, and delisting policies. Read this after you have gone through the main registry overview and authentication guides, since many answers assume familiarity with those workflows. The sections on troubleshooting publishing failures are particularly valuable if you encounter cryptic errors during your first submission attempt.

Github Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘GithubSearchTool’ is designed to search websites and convert them into clean markdown or structured data.

Google Serper Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘SerperDevTool’ is designed to search the internet and return the most relevant results.

Governance and Stewardship

Mon, 01 Jan 0001 00:00:00 +0000

Learn about the Model Context Protocol’s governance structure and how to participate in the community

Group Charter Template

Mon, 01 Jan 0001 00:00:00 +0000

Template for MCP Working Group and Interest Group charters.

How to Authenticate When Publishing to the Official MCP Registry

Mon, 01 Jan 0001 00:00:00 +0000

Authentication is the first barrier you will hit when trying to publish an MCP server to the official registry, and getting it wrong means silent failures with unhelpful error messages. Focus on how GitHub-based identity verification works and the relationship between your GitHub account and registry publishing permissions. If you are building CI/CD pipelines for MCP server releases, read this before the GitHub Actions guide so you understand the credential flow that automation depends on.

How to Automate Publishing with GitHub Actions

Mon, 01 Jan 0001 00:00:00 +0000

Automating MCP server publishing through GitHub Actions eliminates manual release steps and ensures every tagged release reaches the registry consistently. The guide walks through workflow file configuration, secret management for registry credentials, and trigger conditions for version-based publishing. A common pitfall is misconfiguring the authentication token scope, which produces permissions errors that surface only at publish time rather than during the workflow setup. Read the authentication guide first to understand the credential requirements before wiring up the action.

Introduction Design Philosophy

Mon, 01 Jan 0001 00:00:00 +0000

JSON RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘JSONSearchTool’ is designed to search JSON files and return the most relevant results.

Key Changes

Mon, 01 Jan 0001 00:00:00 +0000

LangChain overview

Mon, 01 Jan 0001 00:00:00 +0000

LangChain is an open source framework with a pre-built agent architecture and integrations for any model or tool — so you can build agents that adapt as fast as the ecosystem evolves

LangChain overview

Mon, 01 Jan 0001 00:00:00 +0000

LangChain is an open source framework with a pre-built agent architecture and integrations for any model or tool — so you can build agents that adapt as fast as the ecosystem evolves

LangGraph overview

Mon, 01 Jan 0001 00:00:00 +0000

Gain control with LangGraph to design agents that reliably handle complex tasks

LangGraph overview

Mon, 01 Jan 0001 00:00:00 +0000

Gain control with LangGraph to design agents that reliably handle complex tasks

LangSmith MCP Server

Mon, 01 Jan 0001 00:00:00 +0000

Use the Model Context Protocol (MCP) server to let language models fetch conversation history, prompts, runs, datasets, experiments, and billing from LangSmith.

Lifecycle

Mon, 01 Jan 0001 00:00:00 +0000

The lifecycle specification defines the three phases every MCP connection goes through: initialization, operation, and shutdown. Focus especially on the initialization handshake where client and server exchange capabilities – getting this wrong is the most common source of connection failures. Note that the protocol version negotiation happens here too, so mismatched versions between client and server will fail fast at this stage rather than producing mysterious errors later.

Logging

Mon, 01 Jan 0001 00:00:00 +0000

Look at Your Data

Mon, 01 Jan 0001 00:00:00 +0000

Design your collection schema and chunking strategy based on your data.

MCP

Mon, 01 Jan 0001 00:00:00 +0000

MCP is an open standard protocol for seamless AI model integration with data sources and tools

MCP Apps

Mon, 01 Jan 0001 00:00:00 +0000

Interactive UI applications that render inside MCP hosts like Claude Desktop

MCP Inspector

Mon, 01 Jan 0001 00:00:00 +0000

In-depth guide to using the MCP Inspector for testing and debugging Model Context Protocol servers

MCP Registry Aggregators

Mon, 01 Jan 0001 00:00:00 +0000

MCP Registry Supported Package Types

Mon, 01 Jan 0001 00:00:00 +0000

The MCP registry supports multiple package types and choosing the wrong one constrains how clients can discover and install your server. Understanding the distinction between npm packages, Docker images, and other supported formats is critical before you publish, because changing package type after initial publication creates migration headaches for existing consumers. Focus on the compatibility matrix between package types and host applications, since not every MCP client supports every format equally well.

MCP Tools

Mon, 01 Jan 0001 00:00:00 +0000

Load additional tools from MCP (Model Context Protocol) servers

MCP, A2A, and AG-UI

Mon, 01 Jan 0001 00:00:00 +0000

Understanding how AG-UI complements and works with MCP and A2A

MDX RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘MDXSearchTool’ is designed to search MDX files and return the most relevant results.

Middleware

Mon, 01 Jan 0001 00:00:00 +0000

Connect to existing protocols, in process agents or custom solutions via AG-UI

Model Context Protocol

Mon, 01 Jan 0001 00:00:00 +0000

Connect MCP servers so agents can request external data or actions through standardized tool APIs.

Model Context Protocol (MCP)

Mon, 01 Jan 0001 00:00:00 +0000

Model Context Protocol (MCP)

Mon, 01 Jan 0001 00:00:00 +0000

Model Context Protocol (MCP)

Mon, 01 Jan 0001 00:00:00 +0000

LangChain’s MCP integration allows you to expose MCP servers as LangChain tools, bridging two of the most important ecosystems in the AI tooling space. This is particularly valuable if you have existing MCP servers and want to use them within LangChain agents or LangGraph workflows without rewriting tool definitions. Focus on how MCP tool schemas map to LangChain’s tool interface and what metadata is preserved or lost in translation. Be aware that MCP’s streaming and sampling capabilities may not map perfectly to LangChain’s tool abstraction, so test complex MCP servers thoroughly after integration.

Model Context Protocol (MCP) and Weave

Mon, 01 Jan 0001 00:00:00 +0000

Trace activity between your MCP client and MCP server with Weave

MySQL RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘MySQLSearchTool’ is designed to search MySQL databases and return the most relevant results.

NL2SQL Tool

Mon, 01 Jan 0001 00:00:00 +0000

The ‘NL2SQLTool’ is designed to convert natural language to SQL queries.

OAuth Client Credentials

Mon, 01 Jan 0001 00:00:00 +0000

Machine-to-machine authentication for MCP using the OAuth 2.0 client credentials flow

Official MCP Registry Terms of Service

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Draft changes being considered for the AG-UI protocol

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Core concepts in the Agent User Interaction Protocol SDK

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Core concepts in the Agent User Interaction Protocol SDK

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Documentation for encoding Agent User Interaction Protocol events

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Pagination

Mon, 01 Jan 0001 00:00:00 +0000

PDF RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘PDFSearchTool’ is designed to search PDF files and return the most relevant results.

PG RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘PGSearchTool’ is designed to search PostgreSQL databases and return the most relevant results.

Pinecone Assistant architecture

Mon, 01 Jan 0001 00:00:00 +0000

Ping

Mon, 01 Jan 0001 00:00:00 +0000

Platform Overview

Mon, 01 Jan 0001 00:00:00 +0000

Architecture, deployment lifecycle, and core concepts for Dedicated Container Inference.

Production Architecture

Mon, 01 Jan 0001 00:00:00 +0000

Best practices for building production-ready AI applications with CrewAI

Progress

Mon, 01 Jan 0001 00:00:00 +0000

Prompts

Mon, 01 Jan 0001 00:00:00 +0000

Prompts are the least commonly used of the three MCP primitives, but they fill an important niche: reusable, parameterized templates that help users invoke common workflows. Think of them as saved recipes that combine a specific prompt structure with dynamic arguments. Unlike tools and resources, prompts are user-initiated – the user explicitly selects a prompt from a menu rather than the model discovering it automatically. This makes them ideal for standardizing repetitive interactions like “summarize this codebase” or “review this PR.”

Publishing Remote Servers

Mon, 01 Jan 0001 00:00:00 +0000

Remote servers represent a significant architectural shift from the local stdio-based MCP servers most developers start with, enabling cloud-hosted tools accessible over HTTP with persistent sessions. This guide covers the specific requirements for publishing remote servers to the registry, including endpoint URL configuration and health check expectations. Watch for the distinction between the server’s transport URL and its registry metadata URL, as confusing the two is a frequent source of registration failures. Read this after the general registry overview if you plan to deploy MCP servers as hosted services rather than local processes.

Python Guide

Mon, 01 Jan 0001 00:00:00 +0000

Python implementation guide

Python v2 SDK Migration Guide

Mon, 01 Jan 0001 00:00:00 +0000

Migrate from Together Python v1 to v2 - the new Together AI Python SDK with improved type safety and modern architecture.

Quickstart: Publish an MCP Server to the MCP Registry

Mon, 01 Jan 0001 00:00:00 +0000

Publishing your MCP server to the registry makes it discoverable by any MCP client, which is a critical step for adoption beyond your own projects. Focus on the metadata requirements (name, description, capabilities) because poorly described servers are effectively invisible in the registry. Note that the registry currently requires GitHub-based authentication and a specific package manifest format — read through the validation errors carefully if your first publish attempt fails, as the error messages are quite specific. Come to this doc after you have a working server from the build-server tutorial, since you need a functional server before publishing makes sense.

Reference Architecture

Mon, 01 Jan 0001 00:00:00 +0000

W&B Reference Architecture

Reference architectures

Mon, 01 Jan 0001 00:00:00 +0000

Resources

Mon, 01 Jan 0001 00:00:00 +0000

Resources are how MCP servers expose data to the model without requiring the model to “call” anything – think of them as files or documents the application can pull into context. The key distinction from tools is that resources are application-controlled (the host decides when to read them), while tools are model-controlled (the model decides when to invoke them). Pay attention to the URI-based addressing scheme, which lets clients discover and subscribe to resource updates. If your use case is primarily about providing context rather than performing actions, resources are usually the better primitive.

Roadmap

Mon, 01 Jan 0001 00:00:00 +0000

Our plans for evolving Agent User Interaction Protocol

Roadmap

Mon, 01 Jan 0001 00:00:00 +0000

Our plans for evolving Model Context Protocol

Roots

Mon, 01 Jan 0001 00:00:00 +0000

Sample Methods

Mon, 01 Jan 0001 00:00:00 +0000

Comprehensive examples of protocol methods and responses

Sample Methods

Mon, 01 Jan 0001 00:00:00 +0000

Comprehensive examples

Sampling

Mon, 01 Jan 0001 00:00:00 +0000

Sampling is the mechanism that allows an MCP server to request LLM completions through the client, effectively enabling servers to leverage AI capabilities without embedding their own model access. This is one of the most powerful and least intuitive parts of the MCP specification because it inverts the typical client-server relationship. Pay careful attention to the human-in-the-loop requirements, since the spec mandates that clients must obtain user approval before fulfilling sampling requests, which has significant UX implications. If you are building agentic MCP servers that need to reason or generate text, understanding this capability is essential.

Schema Reference

Mon, 01 Jan 0001 00:00:00 +0000

Scrape Website

Mon, 01 Jan 0001 00:00:00 +0000

The ‘ScrapeWebsiteTool’ is designed to extract and read the content of a specified website.

SDK Tiering System

Mon, 01 Jan 0001 00:00:00 +0000

Feature completeness, protocol support, and maintenance commitment levels for Model Context Protocol SDKs

SDKs

Mon, 01 Jan 0001 00:00:00 +0000

Official SDKs for building with Model Context Protocol

Security Best Practices

Mon, 01 Jan 0001 00:00:00 +0000

Security considerations, attack vectors, and best practices for MCP implementations

Selenium Scraper

Mon, 01 Jan 0001 00:00:00 +0000

The ‘SeleniumScrapingTool’ is designed to extract and read the content of a specified website using Selenium.

SEP Guidelines

Mon, 01 Jan 0001 00:00:00 +0000

Specification Enhancement Proposal (SEP) guidelines for proposing changes to the Model Context Protocol

SEP-1024: MCP Client Security Requirements for Local Server Installation

Mon, 01 Jan 0001 00:00:00 +0000

MCP Client Security Requirements for Local Server Installation

SEP-1024: MCP Client Security Requirements for Local Server Installation

Mon, 01 Jan 0001 00:00:00 +0000

MCP Client Security Requirements for Local Server Installation

SEP-1034: Support default values for all primitive types in elicitation schemas

Mon, 01 Jan 0001 00:00:00 +0000

Support default values for all primitive types in elicitation schemas

SEP-1034: Support default values for all primitive types in elicitation schemas

Mon, 01 Jan 0001 00:00:00 +0000

Support default values for all primitive types in elicitation schemas

SEP-1036: URL Mode Elicitation for secure out-of-band interactions

Mon, 01 Jan 0001 00:00:00 +0000

URL Mode Elicitation for secure out-of-band interactions

SEP-1036: URL Mode Elicitation for secure out-of-band interactions

Mon, 01 Jan 0001 00:00:00 +0000

URL Mode Elicitation for secure out-of-band interactions

SEP-1046: Support OAuth client credentials flow in authorization

Mon, 01 Jan 0001 00:00:00 +0000

Support OAuth client credentials flow in authorization

SEP-1046: Support OAuth client credentials flow in authorization

Mon, 01 Jan 0001 00:00:00 +0000

Support OAuth client credentials flow in authorization

SEP-1302: Formalize Working Groups and Interest Groups in MCP Governance

Mon, 01 Jan 0001 00:00:00 +0000

Formalize Working Groups and Interest Groups in MCP Governance

SEP-1302: Formalize Working Groups and Interest Groups in MCP Governance

Mon, 01 Jan 0001 00:00:00 +0000

Formalize Working Groups and Interest Groups in MCP Governance

SEP-1303: Input Validation Errors as Tool Execution Errors

Mon, 01 Jan 0001 00:00:00 +0000

Input Validation Errors as Tool Execution Errors

SEP-1303: Input Validation Errors as Tool Execution Errors

Mon, 01 Jan 0001 00:00:00 +0000

Input Validation Errors as Tool Execution Errors

SEP-1319: Decouple Request Payload from RPC Methods Definition

Mon, 01 Jan 0001 00:00:00 +0000

Decouple Request Payload from RPC Methods Definition

SEP-1319: Decouple Request Payload from RPC Methods Definition

Mon, 01 Jan 0001 00:00:00 +0000

Decouple Request Payload from RPC Methods Definition

SEP-1330: Elicitation Enum Schema Improvements and Standards Compliance

Mon, 01 Jan 0001 00:00:00 +0000

Elicitation Enum Schema Improvements and Standards Compliance

SEP-1330: Elicitation Enum Schema Improvements and Standards Compliance

Mon, 01 Jan 0001 00:00:00 +0000

Elicitation Enum Schema Improvements and Standards Compliance

SEP-1577: Sampling With Tools

Mon, 01 Jan 0001 00:00:00 +0000

Sampling With Tools

SEP-1577: Sampling With Tools

Mon, 01 Jan 0001 00:00:00 +0000

Sampling With Tools

SEP-1613: Establish JSON Schema 2020-12 as Default Dialect for MCP

Mon, 01 Jan 0001 00:00:00 +0000

Establish JSON Schema 2020-12 as Default Dialect for MCP

SEP-1613: Establish JSON Schema 2020-12 as Default Dialect for MCP

Mon, 01 Jan 0001 00:00:00 +0000

Establish JSON Schema 2020-12 as Default Dialect for MCP

SEP-1686: Tasks

Mon, 01 Jan 0001 00:00:00 +0000

Tasks

SEP-1686: Tasks

Mon, 01 Jan 0001 00:00:00 +0000

Tasks

SEP-1699: Support SSE polling via server-side disconnect

Mon, 01 Jan 0001 00:00:00 +0000

Support SSE polling via server-side disconnect

SEP-1699: Support SSE polling via server-side disconnect

Mon, 01 Jan 0001 00:00:00 +0000

Support SSE polling via server-side disconnect

SEP-1730: SDKs Tiering System

Mon, 01 Jan 0001 00:00:00 +0000

SDKs Tiering System

SEP-1730: SDKs Tiering System

Mon, 01 Jan 0001 00:00:00 +0000

SDKs Tiering System

SEP-1850: PR-Based SEP Workflow

Mon, 01 Jan 0001 00:00:00 +0000

PR-Based SEP Workflow

SEP-1850: PR-Based SEP Workflow

Mon, 01 Jan 0001 00:00:00 +0000

PR-Based SEP Workflow

SEP-1865: MCP Apps - Interactive User Interfaces for MCP

Mon, 01 Jan 0001 00:00:00 +0000

MCP Apps - Interactive User Interfaces for MCP

SEP-1865: MCP Apps - Interactive User Interfaces for MCP

Mon, 01 Jan 0001 00:00:00 +0000

MCP Apps - Interactive User Interfaces for MCP

SEP-2085: Governance Succession and Amendment Procedures

Mon, 01 Jan 0001 00:00:00 +0000

Governance Succession and Amendment Procedures

SEP-2085: Governance Succession and Amendment Procedures

Mon, 01 Jan 0001 00:00:00 +0000

Governance Succession and Amendment Procedures

SEP-2133: Extensions

Mon, 01 Jan 0001 00:00:00 +0000

Extensions

SEP-2133: Extensions

Mon, 01 Jan 0001 00:00:00 +0000

Extensions

SEP-2148: MCP Contributor Ladder

Mon, 01 Jan 0001 00:00:00 +0000

MCP Contributor Ladder

SEP-2149: MCP Group Governance and Charter Template

Mon, 01 Jan 0001 00:00:00 +0000

MCP Group Governance and Charter Template

SEP-2207: OIDC-Flavored Refresh Token Guidance

Mon, 01 Jan 0001 00:00:00 +0000

OIDC-Flavored Refresh Token Guidance

SEP-2260: Require Server requests to be associated with a Client request.

Mon, 01 Jan 0001 00:00:00 +0000

Require Server requests to be associated with a Client request.

SEP-414: Document OpenTelemetry Trace Context Propagation Conventions

Mon, 01 Jan 0001 00:00:00 +0000

Document OpenTelemetry Trace Context Propagation Conventions

SEP-414: Document OpenTelemetry Trace Context Propagation Conventions

Mon, 01 Jan 0001 00:00:00 +0000

Document OpenTelemetry Trace Context Propagation Conventions

SEP-932: Model Context Protocol Governance

Mon, 01 Jan 0001 00:00:00 +0000

Model Context Protocol Governance

SEP-932: Model Context Protocol Governance

Mon, 01 Jan 0001 00:00:00 +0000

Model Context Protocol Governance

SEP-973: Expose additional metadata for Implementations, Resources, Tools and Prompts

Mon, 01 Jan 0001 00:00:00 +0000

Expose additional metadata for Implementations, Resources, Tools and Prompts

SEP-973: Expose additional metadata for Implementations, Resources, Tools and Prompts

Mon, 01 Jan 0001 00:00:00 +0000

Expose additional metadata for Implementations, Resources, Tools and Prompts

SEP-985: Align OAuth 2.0 Protected Resource Metadata with RFC 9728

Mon, 01 Jan 0001 00:00:00 +0000

Align OAuth 2.0 Protected Resource Metadata with RFC 9728

SEP-985: Align OAuth 2.0 Protected Resource Metadata with RFC 9728

Mon, 01 Jan 0001 00:00:00 +0000

Align OAuth 2.0 Protected Resource Metadata with RFC 9728

SEP-986: Specify Format for Tool Names

Mon, 01 Jan 0001 00:00:00 +0000

Specify Format for Tool Names

SEP-986: Specify Format for Tool Names

Mon, 01 Jan 0001 00:00:00 +0000

Specify Format for Tool Names

SEP-990: Enable enterprise IdP policy controls during MCP OAuth flows

Mon, 01 Jan 0001 00:00:00 +0000

Enable enterprise IdP policy controls during MCP OAuth flows

SEP-990: Enable enterprise IdP policy controls during MCP OAuth flows

Mon, 01 Jan 0001 00:00:00 +0000

Enable enterprise IdP policy controls during MCP OAuth flows

SEP-991: Enable URL-based Client Registration using OAuth Client ID Metadata Documents

Mon, 01 Jan 0001 00:00:00 +0000

Enable URL-based Client Registration using OAuth Client ID Metadata Documents

SEP-991: Enable URL-based Client Registration using OAuth Client ID Metadata Documents

Mon, 01 Jan 0001 00:00:00 +0000

Enable URL-based Client Registration using OAuth Client ID Metadata Documents

SEP-994: Shared Communication Practices/Guidelines

Mon, 01 Jan 0001 00:00:00 +0000

Shared Communication Practices/Guidelines

SEP-994: Shared Communication Practices/Guidelines

Mon, 01 Jan 0001 00:00:00 +0000

Shared Communication Practices/Guidelines

Server Card Charter

Mon, 01 Jan 0001 00:00:00 +0000

Charter for the MCP Server Card Working Group.

Specification

Mon, 01 Jan 0001 00:00:00 +0000

Complete A2A protocol specifications

Specification

Mon, 01 Jan 0001 00:00:00 +0000

Specification Enhancement Proposals (SEPs)

Mon, 01 Jan 0001 00:00:00 +0000

Index of all MCP Specification Enhancement Proposals

Specification Enhancement Proposals (SEPs)

Mon, 01 Jan 0001 00:00:00 +0000

Index of all MCP Specification Enhancement Proposals

Spider Scraper

Mon, 01 Jan 0001 00:00:00 +0000

The ‘SpiderTool’ is designed to extract and read the content of a specified website using Spider.

Tasks

Mon, 01 Jan 0001 00:00:00 +0000

Test Agent Card

Mon, 01 Jan 0001 00:00:00 +0000

Online tool to validate if a domain supports the A2A protocol and visualize agent card information. Enter any URL to check for A2A protocol support and parse the agent.json file.

The MCP Registry

Mon, 01 Jan 0001 00:00:00 +0000

This is the starting point for understanding the MCP registry, the centralized discovery layer that lets host applications find and install MCP servers by name. The registry is what turns MCP from a protocol into an ecosystem, so grasping its design goals and trust model is important before publishing or consuming servers. Read this overview first, then move to authentication and package types for the operational details. If you are evaluating whether to publish your MCP server publicly or keep it private, this page covers the discoverability and governance tradeoffs.

The MCP Registry Moderation Policy

Mon, 01 Jan 0001 00:00:00 +0000

Tools

Mon, 01 Jan 0001 00:00:00 +0000

Tools are the most commonly used MCP primitive – they let the model invoke server-side functions with structured inputs and receive results. Focus on the JSON Schema-based input validation, which is how the model knows what arguments a tool accepts. A critical detail is that tool calls are model-initiated but require human approval in most host implementations, so design your tool descriptions to be clear enough that users understand what they are approving. Keep tool names concise and descriptions precise, as the model relies heavily on them for deciding when and how to call each tool.

Training Overview

Mon, 01 Jan 0001 00:00:00 +0000

Launch RFT jobs using the eval-protocol CLI

Transports

Mon, 01 Jan 0001 00:00:00 +0000

MCP defines two transport mechanisms: stdio (standard input/output) and HTTP with Server-Sent Events (SSE). For local development and CLI-based servers, stdio is simpler and the right default choice. HTTP+SSE is needed when the server runs remotely or must handle multiple clients. A common gotcha is that the Streamable HTTP transport replaced the older SSE-only transport in the 2025-11-25 spec revision, so be careful not to follow outdated examples that use the previous approach.

Triggers and Events Charter

Mon, 01 Jan 0001 00:00:00 +0000

Charter for the MCP Triggers and Events Working Group.

TXT RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘TXTSearchTool’ is designed to perform a RAG (Retrieval-Augmented Generation) search within the content of a text file.

Types

Mon, 01 Jan 0001 00:00:00 +0000

Documentation for the core types used in the Agent User Interaction Protocol SDK

Types

Mon, 01 Jan 0001 00:00:00 +0000

Documentation for the core types used in the Agent User Interaction Protocol Python SDK

TypeScript Guide

Mon, 01 Jan 0001 00:00:00 +0000

TypeScript implementation guide

Understanding A2A

Mon, 01 Jan 0001 00:00:00 +0000

In-depth protocol explanation

Understanding Authorization in MCP

Mon, 01 Jan 0001 00:00:00 +0000

Learn how to implement secure authorization for MCP servers using OAuth 2.1 to protect sensitive resources and operations

Understanding MCP clients

Mon, 01 Jan 0001 00:00:00 +0000

Most developers will build servers rather than clients, but understanding the client side is critical for grasping the full protocol flow. The key insight is that the client acts as a gatekeeper: it decides which server capabilities to expose to the model and enforces security boundaries. Pay attention to how capability negotiation works during initialization, since both sides must agree on what features they support before any real work happens.

Understanding MCP servers

Mon, 01 Jan 0001 00:00:00 +0000

This page explains the three core primitives that MCP servers expose: Tools, Resources, and Prompts. Focus on understanding when to use each one – Tools let the model take actions, Resources provide read-only data the model can pull in, and Prompts are reusable templates for common interactions. A frequent mistake is implementing everything as a tool when a resource would be more appropriate and give the host application more control over how data is presented to the model.

Understanding pod-based indexes

Mon, 01 Jan 0001 00:00:00 +0000

Learn about pod-based index architecture and types

Use an Assistant MCP server

Mon, 01 Jan 0001 00:00:00 +0000

Connect AI agents to Pinecone Assistant via Model Context Protocol.

Versioning

Mon, 01 Jan 0001 00:00:00 +0000

Versioning Published MCP Servers

Mon, 01 Jan 0001 00:00:00 +0000

Getting versioning right is essential for MCP server publishers because clients may pin to specific versions, and breaking changes without a major version bump will silently break downstream integrations. Pay close attention to how semantic versioning interacts with registry resolution rules and what happens when you publish a version that a client cannot upgrade to automatically. If you maintain servers consumed by multiple host applications, read this alongside the package types guide to understand how versioning constraints differ across package formats.

Vision Tool

Mon, 01 Jan 0001 00:00:00 +0000

The ‘VisionTool’ is designed to extract text from images.

Weaviate Vector Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘WeaviateVectorSearchTool’ is designed to search a Weaviate vector database for semantically similar documents using hybrid search.

Website RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘WebsiteSearchTool’ is designed to perform a RAG (Retrieval-Augmented Generation) search within the content of a website.

What is the Model Context Protocol (MCP)?

Mon, 01 Jan 0001 00:00:00 +0000

This is the single best starting point for understanding MCP. Focus on the core value proposition: MCP standardizes how AI applications connect to external data and tools, replacing fragile one-off integrations with a shared protocol. Pay attention to the analogy with USB-C — MCP aims to be a universal connector between LLMs and the systems they need to interact with. The original page also covers building MCP Apps — interactive apps that run inside AI clients — via the MCP Apps overview. Read this before diving into architecture or specification pages.

Working and Interest Groups

Mon, 01 Jan 0001 00:00:00 +0000

Learn about the two forms of collaborative groups within the Model Context Protocol’s governance structure - Working Groups and Interest Groups.

XML RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘XMLSearchTool’ is designed to perform a RAG (Retrieval-Augmented Generation) search within the content of a XML file.

YouTube Channel RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘YoutubeChannelSearchTool’ is designed to perform a RAG (Retrieval-Augmented Generation) search within the content of a Youtube channel.

YouTube Video RAG Search

Mon, 01 Jan 0001 00:00:00 +0000

The ‘YoutubeVideoSearchTool’ is designed to perform a RAG (Retrieval-Augmented Generation) search within the content of a Youtube video.