How to Run an Agent Loop That Actually Converges
A four-checkpoint how-to for agent loops: name a measurable job, bound the tool surface, refresh sandbox fixtures from real traffic, and verify at agent speed.
A four-step path to an agent-ready Astro site: content in an Astro CMS, API/CLI/MCP surfaces generated from one schema, and an edge layer that absorbs spikes while agents call functions instead of clicking.
Astro is a content-first web framework, and EmDash 1.0 shows what a production Astro stack looks like when agents are first-class readers: a CMS with an editor-facing admin plus an API, a CLI, and a built-in MCP server over the same content. The explainer walks four checkpoints: anchor content in the CMS, expose it as API, CLI, and MCP, generate those clients from one API definition through Forge's OpenAPI pipeline, and harden the edge with object caching, a pooled database adapter, and sandboxed plugins. Each checkpoint ends in a concrete outcome, from editors publishing without a deploy to surviving multi-thousand-RPS spikes without taking down the origin.
Agents often work better with a callable command or tool than with a raw REST endpoint. Cloudflare's cf CLI defaults to JSON with condensed formatting to keep token use low, and EmDash ships an API, a CLI, and a built-in MCP server over the same content so an agent can pick the surface that fits its context budget.
No. Forge runs in CI on each API repository, lints every change, and generates preview builds of the CLI, docs, and SDKs from an OpenAPI definition. That keeps new surfaces in sync without a hand-written client for every consumer.
The origin and its database connection pool. Cloudflare migrated its own blog to EmDash and handled millions of pageviews per week with spikes up to 5,000 RPS, which is what drove KV object caching and a Hyperdrive database adapter into the stack.
The posts behind it
Cloudflare introduced cf, a new CLI that provides access to the entire Cloudflare API surface of over 3,000 operations, far exceeding Wrangler’s roughly 280 commands. Cf is designed primarily for AI coding agents, which account for nearly half of Wrangler usage, and defaults to JSON output with condensed formatting to minimize token consumption. The CLI includes a natural language search command, cf cli search, that helps agents discover the right operation without bloating context, and uses TypeScript-based configuration via cloudflare.config.ts for type safety. Cf is generated by Forge, Cloudflare’s internal pipeline that produces CLI commands from OpenAPI schema annotations, enabling consistent and comprehensive API coverage. The open beta is available globally via npm, and cf adopts Vite as its default local development server, with a migration path from Wrangler and maintenance support for 18 months after beta ends.
Cloudflare introduced Forge, an open source, pluggable generation pipeline for producing SDKs, CLIs, documentation, and other API surfaces from API definitions. Forge runs in CI on each team’s API repositories, linting every change and generating preview builds of the CLI, docs, and SDKs with only the changes highlighted, similar to Workers Previews. It supports OpenAPI as input and can chain transformers to produce outputs like Cap’n Web bindings, TanStack Query, or MCP servers, enabling generation of bindings from Workers to other APIs. Forge is designed to reduce coordination overhead across Cloudflare’s hundreds of services and repos, and to enable better API versioning without breaking existing clients or SDKs. It is available under the Apache 2.0 license and can be run for free, privately or publicly, and already generates the output required for the cf CLI.
Kitesurf, Cloudflare’s browser for AI agents built entirely on Workers, has received a major update. It now supports WebMCP, enabling agents to call site functions like searchFlights() instead of simulating clicks, and site owners can enable WebMCP with one switch. The browser passes over 730,000 Web Platform Tests, up 500,000 since launch, and adds support for CSS Layout, CSSOM, custom elements, URL-based module resolution, and improved iframe behavior. Efficiency optimizations reduce DOM traversal overhead, font fetching, and timer work, keeping CPU usage in line with launch benchmarks despite expanded capabilities. Kitesurf also gains full Browser Run API coverage, including Quick Actions from Worker bindings, and a terminal rendering mode using the Kitty protocol or ANSI text, with open-sourcing planned soon.
EmDash 1.0 is a stable, free, and open source content management system built on Astro, offering a production-tested admin for editors and agent-friendly workflows through an API, CLI, and built-in MCP server. Its decentralized plugin registry, built on AT Protocol, lets developers publish plugins without ceding ownership of their identity or releases to a central marketplace, while site owners can discover and install plugins directly. Cloudflare migrated its own blog to EmDash as part of its “Customer Zero” approach, handling millions of pageviews per week and spikes up to 5,000 RPS, which led to features like KV object caching and Hyperdrive database adapter. The project is MIT-licensed with over 175 contributors and translations into 25 languages, and it supports sandboxed plugins for security. EmDash separates the plugin from the catalog, allowing other services to index the same publications without requiring developers to start over.
Cloudflare’s 2026 founders’ letter, marking the company’s 16th birthday, examines rapid changes in internet traffic driven by AI. Since mid-2025, new website creation has exploded, fueled by ‘vibe coding’ tools that allow non-programmers to build apps, with over 7 million developers now using Cloudflare’s platform. Automated traffic from AI agents and crawlers is projected to surpass human traffic by May 2026 and could reach 1,000 times human levels within five years. This surge risks a tragedy of the commons: agents consume resources from many sites but benefit only one, threatening small businesses that lack the information volume older competitors hold. Cloudflare proposes reducing crawler load by fetching only changed content and enabling creators to receive payment when agents use their content, alongside partnerships to shape a fairer internet.
More like this
A four-checkpoint how-to for agent loops: name a measurable job, bound the tool surface, refresh sandbox fixtures from real traffic, and verify at agent speed.
A five-checkpoint walkthrough of BYOC: name the control you need, split control and data planes, build the landing zone, govern data in place, then operate inside your own tenant.
A five-checkpoint workflow for making large-scale training runs bitwise reproducible: pick the guarantee, fingerprint the run, and trace divergence down to the kernel.
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
One useful email. Unsubscribe anytime.