Skills over MCP on the streets of Tokyo

Key takeaways

Skills over MCP transform the MCP server from a tool repository into a dynamic distribution channel capable of delivering workflows, instructions, and knowledge precisely when needed.
Progressive discovery is key to managing context size and improving agent scalability by selectively loading resources.
Standardization is vital but challenging, as the AI and developer community are still exploring the full potential and best practices for skills and MCP extensions.
Conceptualizing skills as personas or sub-agents opens new possibilities for personalized and modular AI behaviors.
Handling dependencies, metadata, and security concerns (like prompt injection) are ongoing challenges that require careful design and standardization.
Future improvements will focus on better discovery patterns, scaling strategies, and balancing flexibility with safety in skill management.

Summary

Introduction to AI Enablement at Nordstrom

The episode begins with Ola Hungerford, Principal Engineer for AI Enablement at Nordstrom, sharing her role supporting AI tools and developer tools. Her work involves enabling teams to effectively utilize AI platforms to improve their workflows and capabilities.

Understanding Skills over MCP and Its Use Cases

The core discussion revolves around what 'skills over MCP' really means. Ola explains that MCP (Model Context Protocol) serves as a distribution channel for tools, workflows, and knowledge, allowing agents to access specific instructions only when needed. This setup helps orchestrate tools, distribute knowledge, and manage memory, making AI interactions more efficient and context-aware. They compare server instructions with skills, emphasizing the benefits of skills for dynamic, conditional, and complex workflows.

Progressive Discovery and Its Benefits

A significant theme is progressive discovery, a mechanism to avoid context explosion by only loading resources when necessary. Ola highlights how this approach applies beyond tools to storing knowledge, preferences, and memory. Skills facilitate this process by providing a standardized, incremental way for agents to fetch resources, leading to more scalable and manageable AI interactions.

Development and Standardization of the MCP Extension

Ola describes the journey from draft status to official release of an MCP extension, driven by the community's interest and similar patterns observed across organizations. She discusses feedback from early adopters, especially regarding how to handle custom metadata and security concerns like prompt injection. The conversation emphasizes the challenge of standardizing features when everyone is still exploring their use cases, maintaining flexibility while avoiding potential abuses.

Skills as Personas and Sub-Agents, Dependency Challenges, and Future Directions

The episode explores the conceptualization of skills as personas or sub-agents, emphasizing the need for standard ways to declare dependencies. Ola admits that current implementations are varied, leading to fragmentation. She talks about ongoing efforts for better standardization, including consent mechanisms, and the importance of balancing flexibility with safety. The discussion ends with a forward-looking view on handling the increasing scale of skills, optimizing discovery patterns, and making agents more efficient and user-friendly.

Show notes

Tool descriptions tell an agent what a tool does. They don't tell it how to use five tools together, in the right order, following your conventions. That gap is where this conversation lives.

Filmed at AGNTCon + MCPCon in Tokyo with Ola Hungerford, Principal Engineer for AI Enablement at Nordstrom and a maintainer of the Model Context Protocol, who spent the last several months turning a pattern everyone was quietly reinventing into an actual MCP extension.

Ola walks through what skills over MCP really means: the server stops being a pile of tools and becomes a distribution channel, handing the agent the instructions, workflows and knowledge it needs only at the moment it needs them. She explains why server instructions weren't enough, how progressive discovery keeps context from exploding, and why the same mechanism works for memory and preferences even when no tools are involved.Then it gets into the harder parts. What belongs in the MCP spec versus the agent skills spec. Why passing custom front matter through opens a rug pull and prompt injection surface nobody wanted. Where skills start to look like sub-agents, and why there's still no standard way to declare which servers a skill depends on. And the honest problem underneath all of it: how do you standardize something while everyone is still finding out what it's actually for, without breaking a hundred things the next time you change your mind?

Timestamps:[0:00] Intro[0:21] AI enablement at Nordstrom[0:33] What skills over MCP actually is[1:29] The MCP server as a distribution channel[2:01] Server instructions versus skills[3:11] Distributing knowledge and memory[4:23] Progressive discovery explained[5:23] Where the idea came from[6:59] From draft to official extension[8:19] What early adopters changed[8:51] Front matter and custom metadata[9:57] Rug pulls and prompt injection risk[10:57] Will any of this get standardized[12:10] Marrying two very different specs[13:03] Skills as personas and sub-agents[13:53] The missing dependency standard[15:15] How the extension actually works[16:19] What harnesses still need to support[16:59] Consent and skill integrity[17:56] Where skills over MCP goes next[18:43] Why cramming 200 tools fails[19:41] Standardizing before you know the answer[22:18] Is git the wrong tool for agents[23:19] Picking tools for the actual persona[23:51] Trying to be less productive[25:48] The anxiety of idle agents[27:15] Why she keeps a robot on her desk[28:22] If the agent feels the friction, does it matter[29:41] Efficiency, waste, and caring enough[30:52] Letting an agent debug for you[32:03] Choosing your rabbit hole