Pinned Tweet
A year ago, I started building on @CloudflareDev. Since then, I’ve had hundreds of requests for a course. I'm excited to announce it's finally ready - The Full-stack on Cloudflare Course
This resonates. After adopting @EffectTS_ I rarely reach for SDKs. I find myself implementing a subset of an SDK with the Effect http client so I can define my own error management logic. This pattern has been especially useful when working with the stripe API
1
22
1,214
If you enjoy @herdrdev and want solid setup for managing remote agents, you can take a look at this repo. I went through the process of setting this up on a Mini PC this morning, and documented process here.
github.com/backpine/remote-a…
Whether you're using Effect or not, documenting the layers of your service like this serves two core purposes:
1. It's really effective at guiding your agent towards a better design.
2. Most importantly it'll help you personally stay on top of your codebase.
my "plans" largely look like pseudo code composed of mostly types/interfaces, how they compose, and their boundaries
ive recently started including call stacks - been very helpful for both me and agents when implementing
1
12
1,048
backpine labs retweeted
Thrilled about this change to Workers.
Each Workers request can now make WAY more subrequests.
Previously, it was capped at 1000, which could be tough if you had long-running Websockets or Workflows.
Not an issue any more...
backpine labs retweeted
Introducing pg_strict for Postgres.
Our new extension adds a safety net to Postgres, catching dangerous queries before they run.
Honest question for small American SaaS companies (<$60k mrr) with customers in the EU selling on Stripe. How have you been handling VAT compliance?
It seems most small companies are just ignoring it.
Replying to @EffectTS_
@EffectTS_ folks — looking for feedback on a pattern I've landed on for Cloudflare Workers.
Workers actively block I/O objects from being shared between requests. A TCP connection opened in one request cannot be used by another.
This breaks the normal Effect pattern where database services are global layers — memoized, type-safe, injected everywhere.
The tension is that I still want ManagedRuntime for the ApiLayer. Router config, middleware, all the static stuff that should be built once and reused. But database connections need to live and die with each request, and those handlers are already bound to the memoized runtime.
My current approach is FiberRef with Effect.locally. At the request boundary, I open a scoped connection and set it on the ref. Handlers read from the ref.
This works but you lose type safety. Nothing forces you to provide the database. Handlers compile fine and die at runtime if you forget.
The alternative is spinning up a fresh ManagedRuntime per request with the database layer included. But rebuilding the full layer graph on every request feels like too much overhead.
Is FiberRef the right call here? Or is there a better pattern for request-scoped services alongside a memoized runtime?
Link to the repo I have been working on. The Readme dives deeper into the problem at hand: github.com/backpine/effect-w…
8
949
A surprisingly good AI workflow I’ve been using for UIs with a lot of data to display:
1. Create an API endpoint that can handle read queries: /api/query.
2. Create a Markdown file explaining, at a high level, the important tables and what the data represents. Also point the agents to a schema file (the Drizzle schema file works fine for this).
3. Write a prompt that kicks off two agents (a UX Agent and a Product Engineer).
The UX Agent is tasked with identifying useful data views for the UI and designing how those views fit together into a cohesive UX.
Before the UX Agent can use a data view, it must first get approval from the Product Engineer.
The Product Engineer writes queries and runs curl commands to validate the data and ensure it makes sense for the proposed use case.
The UX Agent should create a Markdown file documenting:
- the raw queries
- why they are useful
- and how they can be used in a UI
By the end of this process, you’ll likely discover useful features and data views that you hadn’t thought of before.
This file also becomes a reference when designing a more complex UI. You can even use it to instruct another agent to translate the raw queries into ORM queries that live behind well-defined methods.
I have been playing around with this concept on some non-critical workloads. Using @EffectTS_ to build durable steps under the scope of a workflow is really nice.
One thing I patricianly like is you can define durable retries, but catch tagged errors within the scope of the workflow to control when retries actual run.
Let me know if you'd be interested in testing this library as I develop it.
I was originally very reluctant to invest the time into creating a course. Seeing messages like this makes me very happy I did.
5
19
1,283
I've been toying around with this idea (@EffectTS_ + Durable Objects for effectful workflows).
Would you build workflows like this?
Assuming: Each step is durable. Each step can be retried in insolation. Workflows can sleep and resume. You can track the status of all workflow instances globally.
backpine labs retweeted
Our plan to launch super fast.
Big shoutout to @backpinelabs for making us better developers.
Replying to @CloudflareDev
@CloudflareDev now supports Observability Destinations. I recently configured Grafana as a destination and the setup is really cool. If you want to buildout better custom alerting and monitoring for your workers I'd recommend this approach.
So awesome
Big news: Replicate is joining Cloudflare.
Replicate's going to carry on as a distinct brand, and all that will happen is that it’s going to get way better: it’ll be faster, we’ll have more resources, and it’ll integrate with the rest of Cloudflare’s Developer Platform.
replicate.com/blog/replicate…
2
17
1,918
backpine labs retweeted
devs always ask why supabase doesn't offer a Postgres training course
the answer is because we could never make one as well as @aarondfrancis could
Really hope this conference ends up taking place in Utah
I am (hopefully) organizing a conference in the near future for web developers that is uniquely different that most others.
I've been researching and pondering this topic for some time now, so please check it out and convey your interest level!
🔗🧵⬇︎
2
19
3,064
backpine labs retweeted
Recommend anyone want to build real-time feature using Durable Object to watch the video from @backpinelabs. youtube.com/@backpinelabs

























