This week I’m talking with Matt Carey about Code Mode and how most of us have been thinking about MCP all wrong. Matt works on the Agents SDK and MCP at Cloudflare — we discuss how server-side Code Mode lets one MCP server expose all ~2,500 Cloudflare API endpoints in about 1,000 tokens of context, the dynamic Worker loader that runs model-written code safely in a V8 isolate, Matt’s own workflow with Claude, where memory fits into the future of agents, and his Zaggy git wrapper that keeps agents from force-pushing his repos.
Featuring
Sponsors
Coder.com – Secure environments where devs and agents work in parallel. Open by design. Secure by default.
Tailscale – Adam loves Tailscale! Easy, secure, identity-based access to anything. Tailscale deploys quickly and enables Zero Trust access to any resource on your network. From CI/CD runners across multi-cloud environments, to SaaS tools and infrastructure, Tailscale connects it all, seamlessly.
RWX – CI/CD platform for high velocity teams. When agents help developers write code in minutes, validation becomes your bottleneck. RWX gives agents programmatic control, sub-second cached builds, and semantic outputs they can act on. No commit required. Just iterate until CI passes, then push.
Fly.io – The home of Changelog.com — Deploy your apps close to your users — global Anycast load-balancing, zero-configuration private networking, hardware isolation, and instant WireGuard VPN connections. Push-button deployments that scale to thousands of instances. Check out the speedrun to get started in minutes.
Notes & Links
Featured
- Matt Carey on the Cloudflare blog
- You’ve Been a Bad Agent
- Cloudflare
- Code Mode: give agents an entire API in 1,000 tokens
- Code Mode: the better way to use MCP
- Cloudflare MCP Server
- Cloudflare Agents SDK
- Cloudflare Agents SDK on GitHub
Cloudflare platform
MCP and code-mode references
- Model Context Protocol
- Introducing the Model Context Protocol
- Code execution with MCP
- CodeAct paper
- GitHub MCP Server
- Datadog MCP Server
- Pydantic Monty
Coding agents and tools
Agent memory and personal AI
Homelab and infrastructure
Chapters
| Chapter Number | Chapter Start Time | Chapter Title | Chapter Duration |
| 1 | 00:00 | Welcome to The Changelog | 00:49 |
| 2 | 00:49 | Sponsor: Coder.com | 03:07 |
| 3 | 03:56 | It's Matt Carey from Cloudflare! | 01:19 |
| 4 | 05:15 | Are we holding MCP wrong? | 03:36 |
| 5 | 08:51 | MCP (Model Context Protocol) explained | 02:43 |
| 6 | 11:34 | The breakthrough with typescript vs tool calling | 03:42 |
| 7 | 15:15 | The idea behind Code Mode | 04:43 |
| 8 | 19:58 | Mapping MCP to a large API via Code Mode | 04:38 |
| 9 | 24:35 | Implementing the search-and-execute pattern in MCP | 03:10 |
| 10 | 27:45 | Matt's workflow with Claude | 08:20 |
| 11 | 36:06 | Balancing innovation with security | 01:35 |
| 12 | 37:41 | Sponsor: Tailscale | 02:35 |
| 13 | 40:16 | Working alone vs working with a team | 03:42 |
| 14 | 43:58 | How do you choose which projects to prioritize? | 04:58 |
| 15 | 48:55 | How are you managing work/life balance? | 01:34 |
| 16 | 50:30 | Life before and after agents | 06:46 |
| 17 | 57:15 | Adam's flow with agents | 05:08 |
| 18 | 1:02:23 | Prompting via voice | 03:20 |
| 19 | 1:05:44 | Devs who are not using AI... | 06:18 |
| 20 | 1:12:02 | Granola is changing note taking | 04:13 |
| 21 | 1:16:15 | Sponsor: RWX | 02:42 |
| 22 | 1:18:56 | Let's talk about memory for agents | 09:47 |
| 23 | 1:28:43 | Taking inspiration from Homelabs | 01:36 |
| 24 | 1:30:19 | Unleashing my agents on my own cloud (Proxmox and Swamp Club) | 05:44 |
| 25 | 1:36:03 | Running local models | 02:53 |
| 26 | 1:38:56 | Teasing DNShole (more) | 03:21 |
| 27 | 1:42:17 | Key indicators of nefarious traffic | 02:35 |
| 28 | 1:44:52 | Getting started with code mode | 04:11 |
| 29 | 1:49:03 | Try this: --agent | 04:35 |
| 30 | 1:53:38 | Closing thoughts and stuff | 01:13 |