Stop scattering LLM SDK/API calls across your codebase. Here is the 2-file rule that fixed mine

I upgraded an LLM SDK and expected a routine version bump. Instead I had to touch 15+ files, fix...

sabato 23 maggio 2026 New tab

1,357 words~6 min read

I upgraded an LLM SDK and expected a routine version bump.

Instead I had to touch 15+ files, fix breaking changes across four providers, and spend the rest of the day hoping I had not missed one. That was the second time it happened. I knew there would be a third.

If you have ever shipped a production LLM system, you probably recognize the smell:

An SDK minor version renames maxTokens to maxOutputTokens and now 15 files break at runtime, not compile time.

Switching one classification task from Claude to a cheaper model means editing import paths and type signatures in business logic.

Stop scattering LLM SDK/API calls across your codebase. Here is the 2-file rule that fixed mine

Stop scattering LLM SDK/API calls across your codebase. Here is the 2-file rule that fixed mine

Related reading

Making LLM Calls Reliable: Retry, Semaphore, Cache, and Batch

I Gave 13 LLMs the Same Codebase and Asked for a Specification. Six Ran on My…

Don't Trust Your LLM's Safety Promises Across Runtimes

LLM 0.32a0 is a major backwards-compatible refactor

Red-Teaming Your LLM Applications: A Practical Guide to Building Guardrails…

I Audited 30 llms.txt Files in the Wild. 5 Anti-Patterns Are Already Forming.

Related reading

Making LLM Calls Reliable: Retry, Semaphore, Cache, and Batch

I Gave 13 LLMs the Same Codebase and Asked for a Specification. Six Ran on My…

Don't Trust Your LLM's Safety Promises Across Runtimes

LLM 0.32a0 is a major backwards-compatible refactor

Red-Teaming Your LLM Applications: A Practical Guide to Building Guardrails…

I Audited 30 llms.txt Files in the Wild. 5 Anti-Patterns Are Already Forming.