← back
Introducing WebMCP: Agents in the Browser — RL Nabors
Takeaway
Existing web infrastructure can expose content and interactions to agents alongside ordinary browser users.
Summary
- Nabors uses a webcomic archive to explore serving humans in browsers, humans inside agents, and agents browsing for users.
- Local stdio MCP servers run as client-spawned processes, while HTTP servers expose web endpoints suitable for serverless hosting.
- Hosted MCP connectors let users connect through a URL instead of configuring local execution commands.
- The example maps website navigation into structured MCP tools and introduces MCP Apps and WebMCP as complementary agent-facing surfaces.
webmcpmcp-transportsagentic-web
Original description
RL Nabors built a comic reader that renders inside Claude. Full panels, navigation, transcript mode, design matched to the original site. No browser tabs. She is reading her own web comic archive entirely through an agent, and it looks like the website. The talk is a case against chat as the permanent UI of agentic software. Chat is to agents what the terminal was to desktop computing: developers love it, everyone else gets the iPhone eventually. MCP apps bundle HTML, CSS, and JavaScript into a single file that agent interfaces render in an iframe, turning any tool response into a real interactive surface. WebMCP goes the other direction: add tool name and description attributes to forms you already have, and browser agents can call your site's functions directly without screenshot parsing or DOM traversal. Both specs exist now. The web platform has been the infinite canvas all along. Speaker info: https://x.com/nearestnabors / nearestnabors https://github.com/rachelnabors Timestamps 0:07 Introduction and Speaker Background 3:28 Motivations for Future-Proofing the Web Comic Site 5:09 Roadmap: MCP, Apps, and WebMCP 6:05 Understanding Transports: STDIO vs. HTTP 8:57 Defining MCP Tools 10:52 The Role of MCP Resources 11:37 The Case Against 'Starfish' (Chat-Only) UI 13:36 MCP Apps: Building Interactive Agent Interfaces 16:34 Best Practices and Gotchas for MCP Apps 17:38 Introducing WebMCP: Agents in the Browser 19:07 Imperative vs. Declarative Models 20:13 WebMCP Demo with the Debugging Extension 21:47 Leveraging Native Browser APIs (Speech, Animation, etc.)