How to Find the Agent Actions a Prompt Injection Can ControlAn agent can't tell your request from text in an email it reads. List every decision made after that read to see what an injection can reach, and design most out.9 min read·theaicommit.hashnode.devOct 28
How to Correct an LLM Judge's Pass Rate With 100 Human LabelsA model that grades your AI feature makes mistakes too, and they leak into the score. Measure how often it is wrong on a hundred outputs a person checked, and you can take its mistakes back out.10 min read·theaicommit.hashnode.dev22KAOct 26
Why Forced JSON Output Is Always Valid and Sometimes WrongStructured output deletes the tokens your schema forbids, so the JSON always parses. When the model has no legal answer it guesses.9 min read·theaicommit.hashnode.dev22KGOct 24
Why Split Prefill and Decode GPUs Lose Time in the Hand-Off, Not the NetworkMoving the KV cache takes 10 ms, but first-token time hit 1.7 s. How vLLM's pull hand-off stalls the decode loop, and how to size decode slots and fix it.10 min read·theaicommit.hashnode.devOct 23
Designing a per-user AI cost meter: idempotency, cache pricing and aliasesNotes from building FiscaLLM, an open-source tool that joins AI cost to subscription revenue3 min read·theaicommit.hashnode.devOct 23
