Skip to content

Knowledge-gap audit

Answer aloud. Mark a gap when you can use a feature but cannot explain its mechanism or trade-off.

  • What exactly does declare(strict_types=1) affect, and from whose call site?
  • When does an array assignment copy its underlying storage?
  • Contrast self, static and $this.
  • Explain covariance and contravariance with a PHP interface.
  • What state survives between requests under FPM? What changes under Octane?
  • What does OPcache cache, and what does it not cache?
  • Trace a request from public/index.php to emitted response.
  • Why can Laravel resolve some classes without an explicit container binding?
  • What is the ordering of service-provider registration and booting, and why?
  • What is actually called when using a facade?
  • Identify three ways an apparently harmless Eloquent loop can overload a database.
  • What state can leak between Octane requests?
  • Which anomaly does each transaction isolation level permit?
  • Why does a queue retry make non-idempotent code dangerous?
  • What guarantee does a database transaction not provide about an external API call?
  • When is an outbox preferable to dispatching an event after commit?
  • State SRP without using the phrase “does one thing.”
  • Distinguish an entity, value object, DTO and aggregate.
  • When does a repository add value over Eloquent, and when is it ceremony?
  • What behavior belongs in a unit test versus a Laravel feature test?
  • Why can high test coverage coexist with low confidence?
  • Distinguish authentication, authorization and tenant isolation.
  • Why is parameterized SQL insufficient to prevent all injection classes?
  • How do you deploy a schema change that old and new code must use concurrently?
  • Why must queue workers be restarted during many Laravel deployments?
  • Which signals distinguish latency caused by PHP, database contention and a downstream service?
  • Distinguish model parameters, tokens, embeddings and context.
  • Explain why temperature zero does not guarantee identical output.
  • Contrast tool calling, MCP, retrieval and a skill.
  • Name the trust boundaries in an agent reading repository and web content.
  • How would you evaluate whether an AI coding workflow is actually helping?