---
title: "Theaicommit (@theaicommit) | Hashnode"
description: "Daily AI labs for engineers. Read the latest articles by Theaicommit on Hashnode."
url: "https://hashnode.com/@theaicommit"
---

@theaicommit

# Theaicommit

@theaicommit · Joined October 2026

Daily AI labs for engineers

### [How to Find the Agent Actions a Prompt Injection Can Control](https://theaicommit.hashnode.dev/how-to-find-the-agent-actions-a-prompt-injection-can-control?utm_source=hashnode&utm_medium=feed)

An 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.dev](https://theaicommit.hashnode.dev/how-to-find-the-agent-actions-a-prompt-injection-can-control?utm_source=hashnode&utm_medium=feed)

[Oct 28](https://hashnode.com/posts/how-to-find-the-agent-actions-a-prompt-injection-can-control/6ae26c0a4cacaa4cf3e5f3d2)

### [How to Correct an LLM Judge's Pass Rate With 100 Human Labels](https://theaicommit.hashnode.dev/how-to-correct-an-llm-judge-s-pass-rate-with-100-human-labels?utm_source=hashnode&utm_medium=feed)

A 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.dev](https://theaicommit.hashnode.dev/how-to-correct-an-llm-judge-s-pass-rate-with-100-human-labels?utm_source=hashnode&utm_medium=feed)

[2](https://hashnode.com/posts/how-to-correct-an-llm-judge-s-pass-rate-with-100-human-labels/6adfe1a711b3cfdff7734fc5) [Oct 26](https://hashnode.com/posts/how-to-correct-an-llm-judge-s-pass-rate-with-100-human-labels/6adfe1a711b3cfdff7734fc5)

### [Why Forced JSON Output Is Always Valid and Sometimes Wrong](https://theaicommit.hashnode.dev/why-forced-json-output-is-always-valid-and-sometimes-wrong?utm_source=hashnode&utm_medium=feed)

Structured 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.dev](https://theaicommit.hashnode.dev/why-forced-json-output-is-always-valid-and-sometimes-wrong?utm_source=hashnode&utm_medium=feed)

[2](https://hashnode.com/posts/why-forced-json-output-is-always-valid-and-sometimes-wrong/6adcc7bb3afe81e3e633f8aa) [Oct 24](https://hashnode.com/posts/why-forced-json-output-is-always-valid-and-sometimes-wrong/6adcc7bb3afe81e3e633f8aa)

### [Why Split Prefill and Decode GPUs Lose Time in the Hand-Off, Not the Network](https://theaicommit.hashnode.dev/why-split-prefill-and-decode-gpus-lose-time-in-the-hand-off-not-the-network?utm_source=hashnode&utm_medium=feed)

Moving 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.dev](https://theaicommit.hashnode.dev/why-split-prefill-and-decode-gpus-lose-time-in-the-hand-off-not-the-network?utm_source=hashnode&utm_medium=feed)

[Oct 23](https://hashnode.com/posts/why-split-prefill-and-decode-gpus-lose-time-in-the-hand-off-not-the-network/6adbaae4c53825a8ef7c7fa3)

### [Designing a per-user AI cost meter: idempotency, cache pricing and aliases](https://theaicommit.hashnode.dev/designing-a-per-user-ai-cost-meter-idempotency-cache-pricing-and-aliases?utm_source=hashnode&utm_medium=feed)

Notes from building FiscaLLM, an open-source tool that joins AI cost to subscription revenue

3 min read · [theaicommit.hashnode.dev](https://theaicommit.hashnode.dev/designing-a-per-user-ai-cost-meter-idempotency-cache-pricing-and-aliases?utm_source=hashnode&utm_medium=feed)

[Oct 23](https://hashnode.com/posts/designing-a-per-user-ai-cost-meter-idempotency-cache-pricing-and-aliases/6adaa4ec28179156a5cecab6)