Model Context Protocol
Synopsis
Model Context Protocol, or MCP, is a standard pattern for connecting AI applications to tools, data, and interactive capabilities through structured servers and clients. In this wiki it also includes MCP Apps and agent-facing interfaces that expose richer actions or UI surfaces to models.
Origin And Context
MCP emerged from the need to standardize how AI clients discover and call tools, access resources, and integrate with external systems. It sits in the lineage of plugin APIs, language-server style tooling, RPC, browser extensions, and developer-tool protocols.
Why It Matters
Agents are only as useful as the tools and context they can safely access. MCP reduces one-off integrations, gives tool providers a common surface, and helps clients reason about capabilities, permissions, and interaction patterns.
How To Use It
Define focused MCP servers with clear tools, schemas, resources, and permission boundaries. Keep tool names concrete, return structured results, test with inspectors, and design for least privilege. For MCP Apps, treat UI and iframe boundaries as part of the security and product contract.
Where It Is Useful
MCP is useful in IDEs, desktop assistants, enterprise data connectors, browser agents, design tools, developer platforms, and internal operations systems.
When To Use It
Use MCP when multiple AI clients need access to the same tools or when a tool provider wants a standard agent-facing integration. For a single narrow app, direct APIs may be simpler until reuse or interoperability matters.
Active Use Cases
- Connecting agents to repositories, browsers, docs, databases, and SaaS tools.
- MCP Apps that return interactive UI from tool servers.
- Agent-ready web and developer-tool integrations.
- Local inspectors and compliance checks for tool servers.
Related Slide Decks
- youtube gcseUQJ6Gbg slides — Using OSS models to build AI apps with millions of users — Hassan El Mghari (4 extracted slide frames)
- youtube aHhB3sjGjkI slides — Agents Building Agents - Alfonso Graziano, Nearform (24 extracted slide frames)
- youtube jVjt 2g8NMY slides — A Genius With Amnesia - Victor Savkin, Nx (19 extracted slide frames)
Related Scheduled Sessions
- 2026 07 01 jan curn mcp doesn t suck your agent does — MCP doesn’t suck — your agent does; Jan Curn (Day 4 — Session Day 3 · 1:55pm-2:15pm · Expo Stage 2 NW; official schedule)
- 2026 06 30 liad yosef mcp apps extending the frontier — MCP Apps - Extending the frontier; Liad Yosef, Ido Salomon (Day 3 — Session Day 2 · 2:25pm-2:45pm · Context Engineering; official schedule)
- 2026 06 30 dustin mihalik mcp apps give the model data give the user a ui — MCP Apps: Give the Model Data, Give the User a UI; Dustin Mihalik (Day 3 — Session Day 2 · 2:50pm-3:10pm · Context Engineering; official schedule)
- 2026 06 29 kwindla kramer the new primitives building ai native software — The New Primitives: Building AI-Native Software; Kwindla Kramer (Day 2 — Session Day 1 · 10:45am-11:05am · Voice & Realtime AI; related YouTube resource; via youtube sAOBXCDiDOs)
- 2026 06 29 pedro lopez how we built the airbyte agent mcp server and cli — How We Built the Airbyte Agent MCP Server and CLI; Pedro Lopez (Day 2 — Session Day 1 · 3:45pm-4:05pm · Expo Stage 1; related YouTube resource; via youtube sAOBXCDiDOs)
- 2026 06 30 joseph wang emulated the data for fully autonomous software engineers and companies — Emulated: The data for fully autonomous software engineers and companies; Joseph Wang (Day 3 — Session Day 2 · 1:55pm-2:15pm · Posttraining & Midtraining; related YouTube resource; via youtube sAOBXCDiDOs)
- 2026 06 30 averi kitsch build time vs run time why your dev tools will fail in production — Build-Time vs. Run-Time: Why Your Dev Tools Will Fail in Production; Averi Kitsch, Prerna Kakkar (Day 3 — Session Day 2 · 10:45am-11:05am · Context Engineering; official schedule)
- 2026 06 29 jim clark who approved that mcp server governing the tool layer — Who Approved That MCP Server? Governing the Tool Layer; Jim Clark (Day 2 — Session Day 1 · 1:55pm-2:15pm · Expo Stage 1 NE; official schedule)
- 2026 06 29 dan adler the enterprise agentic gap when developer level ai tools hit millions of lines — The Enterprise Agentic Gap: When Developer-Level AI Tools Hit Millions of Lines; Dan Adler (Day 2 — Session Day 1 · 10:45am-11:05am · Expo Stage 2 NW; official schedule)
- 2026 06 29 jesse lumarie building the engine while flying the plane launching the figma mcp server — Building the engine while flying the plane — launching the Figma MCP server; Jesse Lumarie (Day 2 — Session Day 1 · 11:10am-11:30am · AI-Native Enterprises; official schedule)
- 2026 06 30 cornelia davis mcp tasks async why the heck aren t any agents supporting mcp tasks async — MCP Tasks (async)/ Why the heck aren't any agents supporting MCP tasks/async?; Cornelia Davis (Day 3 — Session Day 2 · 3:20pm-3:40pm · Context Engineering; official schedule)
- 2026 07 01 sandhya subramani agents that forge their own tools self modifying ai in the wild — Agents That Forge Their Own Tools: Self-Modifying AI in the Wild; Sandhya Subramani (Day 4 — Session Day 3 · 12:05pm-12:25pm · Expo Stage 4 SE; official schedule)
- 2026 06 29 mark lummus burn your flags how paypal designs interactive cli tools for agents — Burn your flags: How PayPal designs interactive CLI tools for agents; Mark Lummus, Navinkumar Patil (Day 1 — Workshop Day · 2:20pm-4:20pm · Workshops Day 1; official schedule)
- 2026 07 01 nikita kothari mcps clis and skills choosing the right tooling layer for agentic development — MCPs, CLIs, and Skills: Choosing the Right Tooling Layer for Agentic Development; Nikita Kothari (Day 4 — Session Day 3 · 11:10am-11:30am · Agentic Engineering; official schedule)
- 2026 06 29 ethan jung min cha dual surface architecture serving humans and agents from the same tool layer — Dual-Surface Architecture: Serving Humans and Agents from the Same Tool Layer; Ethan (Jung Min) Cha (Day 2 — Session Day 1 · 1:55pm-2:15pm · Security; official schedule)
- 2026 06 30 thorsten hans edge native ai building ultra fast agents and mcp servers with spin — Edge-Native AI: Building Ultra-Fast Agents and MCP Servers with Spin; Thorsten Hans (Day 3 — Session Day 2 · 1:55pm-2:15pm · Expo Stage 2; official schedule)
- 2026 06 29 maria bledsoe using ai tools to teach old apps new tricks — Using AI tools to teach old apps new tricks; Maria Bledsoe (Day 2 — Session Day 1 · 2:25pm-2:45pm · Track M; official schedule)
- 2026 07 01 anil nadiminti when ai agents pay and sellers monetize building x402 apps for agentic commerce on aws — When AI Agents Pay and Sellers Monetize: Building x402 Apps for Agentic Commerce on AWS; Anil Nadiminti (Day 4 — Session Day 3 · 11:40am-12:00pm · Agentic Commerce; official schedule)
- 2026 06 29 merve noyan skill issue stop deploying vision language models use them with skills to build e2e vision apps on edge — Skill issue: stop deploying vision language models, use them with Skills to build e2e vision apps on edge; Merve Noyan (Day 2 — Session Day 1 · 11:40am-12:00pm · Vision & OCR; official schedule)
- 2026 06 29 vasuman moza ai tools for forward deployed engineering — AI tools for Forward Deployed Engineering; Vasuman Moza (Day 2 — Session Day 1 · 11:40am-12:00pm · Forward Deployed Engineering; official schedule)
- 2026 06 29 pamela fox get started with models in microsoft foundry to build ai apps — Get Started with Models in Microsoft Foundry to Build AI Apps; Pamela Fox (Day 1 — Workshop Day · 9:00am-10:15am · Track M; official schedule)
- 2026 06 30 ashu joshi deploy agents to users in m365 teams and apps — Deploy agents to users in M365, Teams, and apps; Ashu Joshi (Day 3 — Session Day 2 · 3:20pm-3:40pm · Track M; official schedule)
- 2026 07 01 rafael levi video discovery for agentic world model training — Video Discovery for Agentic World-Model Training; Rafael Levi (Day 2 — Session Day 1 · 2:50pm-3:10pm · Expo Stage 2 NW; related YouTube resource; via youtube btxGmN8RvNU)
- 2026 06 29 du an lightfoot agents that own their inference building production ai agents on dedicated gpus — Agents That Own Their Inference: Building Production AI Agents on Dedicated GPUs; Du'an Lightfoot (Day 1 — Workshop Day · 9:00am-11:00am · Track 7; related YouTube resource; via youtube wFTVEDYVJT0)
Related People
- John Craft
- Laurie Voss
- Pamela Fox
- Sandhya Subramani
- Jason Liu
- John Lindquist
- Liad Yosef
- Charlie Guo
- Frank Coyle
- Christopher Manning
- Kwindla Kramer
- Merve Noyan
- Arun Sekhar
- Elizabeth Fuentes Leone
- Harshul Jain
- Tanmay Sah
- Jan Curn
- Ido Salomon
- Dustin Mihalik
- Pedro Lopez
- Joseph Wang
- Averi Kitsch
- Prerna Kakkar
- Jim Clark
Related Companies
- Microsoft
- OpenAI
- Docker
- Arize AI
- Amazon Web Services
- Hugging Face
- NVIDIA
- MCP Apps
- PayPal
- Towards AI
- Bright Data
- Uber
- Warp
- Weights & Biases by CoreWeave
- egghead.io
- Meta
- Neo4j
Transcript And Resource Support
Transcript-backed resources
- youtube sAOBXCDiDOs — MCP Apps: Primitives, discovery, and the Future of Software - Pietro Zullo, Manufact, Inc
- youtube vh2VGuQ3zhY — The 100-Tool Agent Is a Trap - Sohail Shaikh & Ankush Rastogi, Prosodica
- youtube _xIwFcnHqp4 — Building Interactive UIs in VS Code with MCP Apps — Marlene Mhangami & Liam Hampton, GitHub
- youtube ghJmWQCIHRM — The agent-ready web: Simplify user actions with WebMCP — Tara Agyemang, Google
- youtube _B4Pv9ttFgY — Building Agent Interfaces: Lessons from Chrome DevTools (MCP) for Agents — Michael Hablich, Google
- youtube c 2eEv2ou7Y — Why MCP and ChatGPT Apps Use Double Iframes — Frédéric Barthelet, Alpic
- youtube spNAUEgq_A8 — The Future Is Domain-Specific Agents - Justin Schroeder, StandardAgents
- youtube hCMrEfPG2Yg — Beyond Components: Designing Generative UI for MCP Apps — Ruben Casas, Postman
- youtube 4uFVSLgD2Q4 — Agents in Production: How OpenGov Built and Scaled OG Assist - Gabe De Mesa, OpenGov
- youtube btxGmN8RvNU — Your Agent's Biggest Lie: "I Searched the Web" — Rafael Levi, Bright Data
- youtube wFTVEDYVJT0 — Building Agents with Amazon Nova Act and MCP - Du'An Lightfoot, Amazon (Full Workshop)
- youtube x5GEVnkuRw — Structuring the Unstructured - Cedric Clyburn, Red Hat
- youtube SKDJo2CopRs — Why Eval++ Is the Next Great Compute Primitive — Sunil Pai & Matt Carey, Cloudflare
- youtube aHhB3sjGjkI — Agents Building Agents - Alfonso Graziano, Nearform
- youtube 2IxD9OB3XuQ — Continual Learning for AI Agents: From Failures to Durable Improvements - Soheil Feizi, RELAI
- youtube dRmWYHuIJxM — We Cut 94% of AI Coding Tokens With a Local Code Index - Rajkumar Sakthivel, Tesco
- youtube uiP88SpCi1Q — Your Agent Is Wasting Tokens and You Don't Know It - Erik Hanchett, AWS
- youtube bk0TmxoZlUY — Evals 101 — Doug Guthrie, Braintrust
Quote signals
- “I'm an engineer here at OpenGov and today we're going to be talking about agents in production, specifically how OpenGov built and scaled OG Assist.” — youtube 4uFVSLgD2Q4
- “So, that's kind of how we handled long context and memory, and it's worked pretty well for us.” — youtube 4uFVSLgD2Q4
- “Uh it's an open source inspector, again, uh something comparable to the official inspector from the model context protocol maintainers that allows you to test these MCP servers and apps specifically uh on your local machine.” — youtube sAOBXCDiDOs
- “MCP apps is now the official extension of the Model Context Protocol that allows to return UI elements within MCP servers.” — youtube sAOBXCDiDOs
- “[clears throat] And also Cursor allows you to submit their MCP server set uh The way you submit is different for all three, but basically what happens is that you need to make sure that your MCP app is compliant.” — youtube sAOBXCDiDOs
- “The let's say past few decades, we have been building the web for human actions and human eyes, and we've been trying to optimize for that, but these days, it's not just humans that are using the web.” — youtube ghJmWQCIHRM
- “Well, today I'm going to share four engineering lessons from the Chrome uh DevTools team on how we build Chrome DevTools for agents and how we deployed it for effect.” — youtube _B4Pv9ttFgY
- “In a world where you are delegating away work to agents and automating away agents, you need to think about trust boundaries.” — youtube _B4Pv9ttFgY