Should I invent my own HTTP status codes?
Almost never. Clients, proxies, caches and monitoring tools key off the standard classes: 2xx, 3xx, 4xx, 5xx. An unregistered code is treated by its class at best and confuses tooling at worst. Use the closest standard code and put the detail in a structured error body.
Answered in
The Strange HTTP Status Codes: 451, 418, 420, 218 and 530, and What They TeachFive odd status codes, which are standard, which are jokes and which are vendor inventions, and why custom codes are a trap for client code.
Read the full analysisOther questions this article answers
More development best practices questions
- What is the primary security vulnerability of storing JWTs in localStorage?
- How do server-side sessions scale horizontally across multiple instances?
- Why is immediate revocation nearly impossible with stateless JWTs?
- What is the recommended modern authentication architecture for SaaS apps?
- What is the N+1 query problem in GraphQL and how is it fixed?
- How does GraphQL Federation allow multiple backend teams to maintain one API?
- Why is caching harder in GraphQL than in traditional REST APIs?
- When should an engineering team prefer REST or gRPC over GraphQL?
Every answer on Crashtech is written by the editor of the article it comes from — never auto-summarised. Browse all answers or the Development Best Practices beat.