Why Projections Exist — Your First Read Model
https://medium.com/@dariuszgafka/why-projections-exist-your-first-read-model-ceaab247bc2b

https://redd.it/1slz17q
@r_php

Is Claude my permanent co-author?

I wanted to migrate an old PHP web app that I wrote by hand to a modern framework, and chose Symfony. I prepared some docs, watched some symfony youtubes, and resisted getting started for months. Finally, I decided to see if Claude code could get me over the hump. Well, I'm astounded by the result. Completely rebuilt in a solid Symfony framework in about 10 days. Works beautifully. I had claude build documentation as well, but now I have a site whose internal wiring is really beyond my ability to manage responsibly. I can invoke Claude in the code base, and pick up work at any time, but I couldn't maintain the system without Claude. I feel peculiar about it now: I'm the (human) author but I have an AI partner that has to be part of the "team" going forward. I can't be the first person to get here. Any words of advice?

https://redd.it/1slxyp7
@r_php

How I evolved a PHP payment model from one table to DDD — channels, state machines, and hexagonal architecture

I got tired of every project reinventing the payment layer from scratch, so I tried to build a proper domain model in PHP and document the process.

Wrote about going from a single table to channels, state machines, and hexagonal architecture.

It's an experiment, not a final answer — curious how others tackle this.

https://corner4.dev/reinventing-payment-how-i-evolved-a-domain-model-from-one-table-to-ddd

https://redd.it/1sm3f80
@r_php

Spatie Guidelines as AI Skills
https://spatie.be/blog/spatie-guidelines-as-ai-skills

https://redd.it/1sm7v1h
@r_php

SymfonyLive Berlin 2026: "Specing out teamwork”
https://symfony.com/blog/symfonylive-berlin-2026-specing-out-teamwork?utm_medium=feed&utm_source=Symfony%20Blog%20Feed

https://redd.it/1sm8ctw
@r_php

Your AI Agent Has Amnesia. Let's Fix It. - Ship AI with Laravel EP4
https://youtu.be/mZbyCIOFWsE

https://redd.it/1sm8hp0
@r_php

Anyone else get tired of rebuilding Filament resources every time admin requirements change?

I kept hitting the same pattern in Laravel / Filament projects:

the first version of the admin panel is usually fine, but later the data side keeps changing.

New content type.
More custom fields.
Better filtering.
Dashboards.
API requirements.
Tenant-specific behavior.
More exceptions.

At that point, every "small" change becomes another migration, another model, another Filament resource, and another layer of maintenance.

So I built a plugin called **Filament Studio** for Filament v5.

The idea is to let you create collections and fields at runtime, manage records through generated Filament UI, build dashboards, add advanced filtering, and expose APIs without rebuilding a brand-new schema layer every time requirements shift.

It also includes things I thought were important if this is going to be useful beyond a demo:

\- authorization
\- multi-tenancy
\- versioning
\- soft deletes
\- custom field and panel extensibility

I know some people will immediately look at the EAV angle and prefer hand-built resources anyway, which is fair.

I am mostly curious about where other Laravel developers draw that line.

If you are building something with a stable schema, I still think hand-built resources make sense.

But if the admin/data model changes constantly, would you rather keep building each resource manually, or use something like this?

Repo if you want to look at it:

GitHub: https://github.com/flexpik/filament-studio

I am not looking for empty promotion here. I would rather hear the real objections or the kinds of projects where this would actually help.

https://redd.it/1smipbx
@r_php

I built a CLI that generate Symfony-compatible Twig templates from an Astro frontend project

Hello,

Sharing a tool I built for a workflow that comes up often on agency projects :

Astro frontend, Symfony backend, someone has to bridge the two.

Frontmatter Solo reads a constrained Astro project and generates Twig templates structured for Symfony:

frontmatter solo:build --adapter twig

Output drops directly into templates/:

The data contract follows the fm namespace. Your controller passes it as a single array.

INTEGRATION.md is generated automatically, it documents every Twig variable expected by each template. The Symfony developer reads it once, writes the controller, done.

Compatible with Symfony 5, 6, 7 (Twig 3.x).

Also works with Drupal and Craft CMS.

Check compatibility before buying (free):

npx @withfrontmatter/solo-check

$49 one-time https://www.frontmatter.tech/solo/symfony

Happy to discuss the fm namespace convention or the controller integration pattern.

https://redd.it/1smtnqq
@r_php

Introducing TypeScript Transformer 3
https://rubenvanassche.com/introducing-typescript-transformer-3/

https://redd.it/1smw3ek
@r_php

SymfonyLive Berlin 2026: "Simultaneous editing: Easy mode with Symfony UX"
https://symfony.com/blog/symfonylive-berlin-2026-simultaneous-editing-easy-mode-with-symfony-ux?utm_medium=feed&utm_source=Symfony%20Blog%20Feed

https://redd.it/1sn2nmd
@r_php

Two Composer command injection CVEs this week patch is one command, but there's a bigger picture here worth talking about

CVE-2026-40176 (CVSS 7.8) and CVE-2026-40261 (CVSS 8.8), both in Composer's Perforce VCS driver. The 8.8 one fires during `composer install` when pulling from source so your CI pipeline is the actual attack surface. Public PoCs dropped today.

Fix: `composer self-update` to 2.9.6, or 2.2.27 if you're on LTS. One command, do it now.

The reason I'm posting beyond the PSA: this is the third significant PHP ecosystem CVE in about six weeks, and what's getting harder isn't patching it's knowing what you actually need to care about before someone else tells you.

The same vulnerability comes in from four different feeds simultaneously. NVD, GitHub Advisories, OSV, Packagist all reporting the same CVE with different IDs, different severity framings, different context. And CVSS alone tells you very little. These two CVEs have PoCs live today. There are plenty of 9.0+ CVEs with no exploitation evidence that can sit in a backlog forever.

I've been building a tool called A.S.E. to deal with exactly this it watches all the major feeds, deduplicates across them, cross-references against your actual composer.lock, and factors in exploit probability (EPSS) alongside CVSS so you're not just triaging severity theater. Good starting point for anyone wanting something to help them out in these situations

github.com/infinri/A.S.E

https://redd.it/1sn3an5
@r_php

PagibleAI 0.10: PHP CMS for developers AND editors

We just released Pagible 0.10, an open-source AI-powered CMS built as PHP composer package for Laravel applications:

* [https://pagible.com](https://pagible.com)

# What's new in 0.10

* MCP Server — Pagible ships with a built-in Model Context Protocol server. AI agents can create pages, manage content, and search your site programmatically. This makes Pagible one of the first CMS platforms where AI can directly manage your content through a standardized protocol.
* Customizable architecture — The codebase has been split into 9 independent sub-packages (core, admin, AI, GraphQL, search, MCP, theme, etc.). Install only what you need.
* Vuetify 4 admin panel — The admin backend has been upgraded to Vuetify 4 and optimized for WCAG accessibility, keyboard navigation and reduced bundle size.
* Significant performance work — This release focused heavily on database performance: optimized indexes, reduced query count, eager loading, optimized column selection, and faster page tree fetching.
* Rewritten fulltext search — Custom Scout engine supporting fulltext search in SQLite, MySQL/MariaDB, PostgreSQL, and SQL Server. Paginated results with improved relevance ranking.
* Named roles & JSON permissions — Moved from bitmask permissions to a readable JSON array system with configurable roles (e.g. editor, publisher, viewer, etc).
* Security hardening — Rate limiting on all endpoints, strict security, DoS protection against all inputs.

# What makes Pagible different

* API first — GraphQL and JSON:API endpoints out of the box. Build headless sites, mobile apps, or single-page applications without writing a single API route ... or use traditional templates and themes - just as you like.
* AI-native — MCP server for agent-driven content management, plus built-in AI features for content generation, translation, and image manipulation.
* Hierarchical pages — Nested set tree structure with versioning. Editors see drafts, visitors see published content.
* Multi-tenant — Global tenant scoping on all models out of the box.
* Small footprint — The entire codebase is deliberately kept small. No bloat, no unnecessary abstractions.
* LGPL-3.0 — Fully open source.

# Links

* Demo: [https://demo.pagible.com/](https://demo.pagible.com/cmsadmin)
* GitHub: [https://github.com/aimeos/pagible](https://github.com/aimeos/pagible)
* Website: [https://pagible.com](https://pagible.com)

Would love to hear your feedback and if you like it, give a star :-)

https://redd.it/1sn4r04
@r_php

Laravel adds their own product to LLM instructions

https://techstackups.com/articles/laravel-raised-money-and-now-injects-ads-directly-into-your-agent/

https://redd.it/1sn70d7
@r_php

SymfonyLive Berlin 2026: “Git, But Better: An Introduction to Jujutsu (jj)”
https://symfony.com/blog/symfonylive-berlin-2026-git-but-better-an-introduction-to-jujutsu-jj?utm_medium=feed&utm_source=Symfony%20Blog%20Feed

https://redd.it/1sn7jrq
@r_php

Finally moved my PHP media processing to an async Celery (Python) pipeline. Here’s how I handled the cross-language "handshake."

**The Problem:** I was hit with the classic scaling wall: image processing inside request cycles. Doing background removal, resizing, and PDF generation in PHP during a file upload is a recipe for timeouts and a terrible UX. PHP just isn't the right tool for heavy lifting like `rembg` or `ReportLab`.

**The Setup:** I decided to move everything to an async pipeline using **PHP → Redis → Celery (Python) → Cloudinary**.

**The "Aha! 😤 " Moment:** The trickiest part was that PHP doesn't have a great native Celery client. I didn't want to overcomplicate the stack with a bridge, so I just looked at how Celery actually talks to Redis.

Turns out, Celery’s wire format is just JSON. I ended up manually constructing the Celery protocol messages in PHP and pushing them directly into the Redis list. As long as you follow the structure (headers, properties, body), the Python worker picks it up thinking it came from another Celery instance.

**The Pipeline:**

1. **PHP:** Enqueues the job and immediately returns a 202 to the user. No blocking.
2. **Redis:** Acts as the broker.
3. **Celery (Python):** Does the heavy lifting.
* **Background Removal:** `rembg` (absolute lifesaver).
* **Resizing:** `Pillow`.
* **PDFs:** `ReportLab`.
4. **Cloudinary:** Final storage for the processed media.
5. **Callback:** The worker hits a PHP API endpoint to let the app know the asset is ready.

**The Win:** The system is finally snappy. PHP just "enqueues and forgets."

**What I’m fixing in v2:**

* **Dead-letter queues:** Right now, if a job fails, it just logs. I need a better retry/recovery flow.
* **Queue Priority:** Moving heavy PDF tasks to a separate queue so they don't block simple image resizes.
* **Visibility:** Adding **Flower** to actually see what's happening in real-time.
* **Cleanup:** Automating the `/tmp` file purge on the worker side more aggressively.

**Curious if anyone else has gone the "manual protocol" route for cross-language Celery setups?** Is there a cleaner pattern I’m missing, or is this the standard way to bridge the two?

[**https://github.com/eslieh/grid-worker**](https://github.com/eslieh/grid-worker)

[**https://github.com/eslieh/grid**](https://github.com/eslieh/grid)

https://redd.it/1sn1wyh
@r_php

I built a CLI tool that lets your AI agents improve your query performance in a loop
https://redd.it/1sne1ui
@r_php

Sharing Community Feedback from The PHP Foundation

On behalf of The PHP Foundation, I’m excited to share the results of the feedback I’ve collected over the past few weeks. It will help inform The PHP Foundation’s Strategy for the rest of 2026 and into 2027.

There are a lot of opportunities for The PHP Foundation to extend our support into the PHP ecosystem, and I couldn’t be more excited! If you’re interested, you can read the post here:

https://thephp.foundation/blog/2026/04/16/integrating-community-feedback-into-foundation-strategy-part1/

https://redd.it/1snf018
@r_php

Laravel raised money and now injects ads directly into your agent
https://techstackups.com/articles/laravel-raised-money-and-now-injects-ads-directly-into-your-agent/

https://redd.it/1sngr5c
@r_php

PagibleAI 0.10: Laravel CMS for developers AND editors
https://redd.it/1sn4d1i
@r_php

Dependency Injection vs Service Locator
https://slicker.me/php/dependency_injection_vs_service_locator.htm

https://redd.it/1snn2lj
@r_php

Integrating Community Feedback into Foundation Strategy Part 1
https://thephp.foundation/blog/2026/04/16/integrating-community-feedback-into-foundation-strategy-part1/

https://redd.it/1snr71c
@r_php