$ the-wire · showcase
LARAVEL 13 CUTS SQS COSTS WITH BATCH API, AI SDK HITS PHPSTAN LEVEL 2
By RepoJournal · Filed · About Laravel · Composed from the cited sources · methodology
The framework is shipping smarter queue batching that cuts AWS API calls by 95 percent, while the AI package climbs the type-safety ladder and gains a manual testing workbench.
Laravel 13's SQS driver is getting a major efficiency upgrade: the `bulk()` method now dispatches jobs via `SendMessageBatch` instead of looping through individual `SendMessage` calls [1]. A 500-job batch that currently costs 500 API calls will drop to 50 calls or fewer. This hits hard for shops running high-volume work. In parallel, the framework is tidying up deprecated Rector rules and expanding test coverage for relative date query shortcuts [2][3], keeping the codebase modern without ceremony.
Laravel AI just shipped a major stability milestone: full PHPStan level 2 analysis without Larastan or baselines [5]. Every type mismatch was fixed at source, not papered over. That same package landed three heavy-hitting features: a `WithoutBroadcasting` attribute to stop oversized stream events from torpedoing queued jobs [6], an `openai-compatible` driver for pointing at any LLM gateway without inheriting a provider's quirks [7], and a Workbench setup with a live chat UI for real-time testing [8]. The HTTP client is also picking up RFC 10008 support, adding a safe, cacheable `QUERY` method that carries its payload in the request body instead of the URL [4].
Laravel Boost fixed a critical bug where table prefixes were corrupting `ORDER BY DESC` clauses and schema-qualified queries [11], transforming valid SQL into invalid statements. The best-practices skill got a context-saving overhaul, swapping inline rule summaries for a compact index table [9], and enum guideline generation is now deterministic across environments [10].
Action items
- → Upgrade to Laravel 13 for SQS batch improvements if you're running high-volume queued work laravel/framework [plan]
- → Pull Laravel AI's latest if you're using agents with oversized tool results or testing in Workbench laravel/ai [plan]
- → Update Boost immediately if you're using the database-query MCP tool with table prefixes laravel/boost [immediate]
References
- [1] [13.x] Send bulk SQS jobs via SendMessageBatch ↗ laravel/framework
- [2] [13.x] Remove deprecated `StaticCallOnNonStaticToInstanceCallRector` fixer rule ↗ laravel/framework
- [3] [13.x] Add tests for relative date where clauses ↗ laravel/framework
- [4] [13.x] Add `Http::query()` method to the HTTP client ↗ laravel/framework
- [5] Support PHPStan static analysis at level 2 ↗ laravel/ai
- [6] Add WithoutBroadcasting attribute to skip oversized stream events ↗ laravel/ai
- [7] Add configurable OpenAI-compatible provider ↗ laravel/ai
- [8] Add Workbench setup with a chat example for manual testing ↗ laravel/ai
- [9] Restructure laravel-best-practices skill into a compact rule index ↗ laravel/boost
- [10] Make enum guideline generation deterministic ↗ laravel/boost
- [11] Fix table-prefix corrupting "ORDER BY … DESC" and schema-qualified queries (#868) ↗ laravel/boost
