Freek.dev

Anthropic found that Claude's internal reasoning uses a small, limited workspace, not the whole conversation. I break down what that means in practice: stage your prompts one decision at a time instead of dumping a whole feature at once.

Every year I dread coming back from vacation not because of the work, but because I have no idea where I left off. A failing test, a well-written TODO, and a slow first day back are all you need to ease back in.

Agentic commerce is a buzzy term, but is still pretty abstract for developers asking what they can actually build today. This post explains MPP through a working Laravel example, and introduces the new square1/laravel-mpp package.

A practical deep dive into UUIDs, ULIDs, and Sqids, from generation models and sorting behavior to database storage, public IDs, and security trade-offs.

A simple, secure, and fluent PHP API for running untrusted code in isolated Docker containers, talking to the Docker Engine API directly over Guzzle. Sandboxes are locked down by default (no network, non-root, read-only rootfs, resource limits), with fluent managers for containers, images, networks, and volumes.

I updated My popular guide on automated database backups with Docker Compose, now using the renamed nfrastack/container-db-backup image with multi-job support. Now it Includes a handy interactive configurator tool that generates docker-compose and .env files on the fly. Built for devs running multiple Laravel projects who want painless, scheduled backups to S3 or local storage.

CLAUDE.md isn't documentation, it's working memory. A practical approach to keeping it lean and focused instead of letting it grow into an unreadable 300-line dumping ground.

Open source is the substrate the whole industry now runs on — and the quiet engine behind reliable AI coding. Here is why it matters more than ever.

A deep dive into Laravel Context, from request metadata and automatic log enrichment to hidden context, scoped values, queues, scheduled commands, and internals.

A practical guide to stress testing PHP and Laravel applications with VoltTest. It covers PHP-native test scenarios, Laravel Artisan commands, tool comparisons, key metrics like P95/P99 latency, and how to scale tests with VoltTest Cloud.

Replace ugly S3 bucket URLs with clean, cacheable paths through a Laravel proxy. Just 20 lines of controller code, 24-hour CDN caching, and your private bucket stays private.

Wiring a tool onto a Laravel AI agent takes one line, and every tutorial stops there. The hard part is composing several into one loop: scraped web text becomes part of your prompt, tool output quietly balloons your token bill, and async tools never finish in one turn. One real research agent on the Laravel AI SDK toolkit, with the app on GitHub.

Learn how to use Eloquent Query Classes to organize important database logic in Laravel without adding a full repository layer.

Every Laravel RAG tutorial builds the same ingestion pipeline (chunk, embed, store) and stops the moment the agent answers on screen. None of them check whether retrieval is any good. But retrieval quality is decided at ingestion, before the model runs once, and four decisions there fail with no error, no exception, no failed test: - **Chunking** that severs the answer mid-sentence, so `answer@1` falls while `source hit@1` still looks healthy. - An **HNSW index** built with `vector_l2_ops` while you query with cosine `<=>`. Postgres silently ignores the index and scans every row. Laravel 13's native `whereVectorSimilarTo()` hardcodes `<=>`, so it's easier to hit than ever. Shown with `EXPLAIN`. - The **embedding dimension** baked into the `vector(1536)` column type, so "shrink it to save storage" is a migration plus a full re-embed that quietly drops retrieval to 47%. - **Ingesting and querying with different models**, which turns every distance into noise. Each bug is real code from a working repo, proven against an eval suite. It's the prequel to my earlier "Evaluating RAG in Laravel" post: build it, prove it, tune it. Every example verified against `laravel/ai` v0.7.2 and pgvector, with the full repo to clone.

How we use class-based Laravel Pennant features, with a kill switch on every flag and a config-driven path to general availability.

Learn how Expressive can improve a Laravel application by keeping Eloquent as the database layer while moving business logic to fully typed objects.

You can't out-prompt an attacker — to the model, your system instructions and a malicious support ticket are the same text. So stop defending the prompt and lock down the boundaries you actually control: tools scoped to the authenticated user server-side, middleware that screens and logs, output handled as untrusted input, a human in front of anything irreversible, and a fake-free test that fails CI the moment someone drops the auth scope.

Your Agent::fake() tests prove your Laravel AI feature runs — not that its output is any good. This evals a real ticket classifier with the AI SDK: a golden dataset for the fields you can check, an LLM-as-judge for the free text you can't, and a regression gate that catches a bad prompt before your customers do.