Your LLM's JSON passed the schema. It was still wrong.
Deterministic contract tests for structured LLM output on the JVM, and what 120 real documents taught...
Tutorials
Claude Code's Hidden API Costs: Three Ways to Reduce Them
When Claude Code is billed through an API, it is tempting to estimate spend from the visible...
Your AI code reviewer isn't a second opinion. It may be the first opinion again.
A model writes the change. A model reviews the change and says it looks good. The pull request now...
What to Fix When AI Gets Your Brand Wrong
If an AI answer tells a potential customer that your service is unavailable in their area, a more...
Latent State Is the Fields You Never Logged
Giving the heater model one private field and watching it work out a door the logger never recorded. Thirteen weights, gradient descent run backwards through each episode, and the ...
Self-hosting E2B: run E2B code sandboxes on your own servers
Two ways to self-host E2B, and a step-by-step E2B-compatible code sandbox with open-source OpenDsec: one-machine install, code interpreter, fork and snapshots.
The Agent Added a Second Copy of a Utility the Repo Already Had
Why an agent's freshly written helper often duplicates an existing utility, and how silent forks drift into real bugs.
9 Commands That Leave a Coding Agent Waiting for a Keystroke, and the Fix for Each
git commit, npx, apt, ssh, psql, terraform apply, gh auth login: each can stall an AI coding agent on a prompt. The flag or env var that stops each one.
10 Prompts That Turn ChatGPT Into a Senior Data Analyst
10 Prompts That Turn ChatGPT Into a Senior Data Analyst I once spent an entire weekend...
Claude Looked Blocked. My Policy Had Never Fired.
Claude Looked Blocked. My Policy Had Never Fired. From the agentpolicy v0.1.0 release: a...
How to Bulk Download YouTube Transcripts for RAG and AI Summaries (Python)
Getting one YouTube transcript is easy. Getting 500 of them for a RAG pipeline, a research project or...
Open Source Radar · 2026-11-02: Agent infrastructure is splitting into replaceable parts
By Renchao Wu. Originally published on widechaos.cn. For this issue I picked five projects from...
Qwen Image 2.1 Reconstruction: Visual Prompts, Empty Latents, and a Pixel Budget
Understand an image-conditioned generative route without mistaking preservation instructions for fidelity guarantees or a 2K label for fixed dimensions.
QuantDinger Jev Filter Backtest: 0 of 1,324
I rebuilt QuantDinger's Jev trade gate and replayed it on BTC and gold. With defaults it decided 0 of 1,324 entries; without thresholds, a coin flip.
We Open-Sourced Our AI Subtitle Translation Engine: Every SRT Cue In, Every Cue Out
If you've ever pasted an SRT file into an LLM and asked for a translation, you've probably seen this:...
Shipping Ming-Image Design: size the card by its files, and check the alpha in code
A 6B design model peaked at 64.4 GiB on an A100 80GB. Its transparent PNGs depend on a line break, so verify the alpha channel yourself.
How we structure an animated horror reel with Claude Code: timed spec first, then shots, then QA
When you ask Claude Code to "make a 60-second horror reel", it will happily start writing components....
What I Learned Using Free AI Coding Agents on a Real Portfolio (OpenCode and Kilo)
I'm a student, and I built and kept improving my portfolio with free AI coding agents running in the...
What AI code reviewers miss, and the checks I built for it
We replayed 45 findings from a careful human code review against automated rules. They caught 9. Here is what that taught us about which problems tools can own, and the toolkit we ...
Claude Code Custom System Prompts in 2026: Overriding Defaults, Appending Instructions, and What You Cannot Change
Developers discover that Claude Code ignores custom instructions because they set them in the wrong layer. Learn the three ways to shape Claude Code behavior, what the system promp...
Claude Code for Free in VS Code: No Login, No API Key (Ollama + Qwen 3.8)
Last time I ran Claude Code for free in the terminal. A lot of you asked the obvious follow-up: can I...
Building a WhatsApp AI Chatbot That Hands Off to a Human: Architecture and Pitfalls
Getting a language model to answer WhatsApp messages takes an afternoon. Getting it to stop answering...
AI/ML Research Digest — Oct 31, 2026
Accelerated inference & algorithmic efficiency Gradient‑recovery speculative decoding...
The code in AI programming hints that wasn't the student's: half of it ended up in their final submission
Made with AI agents working in my name. Code Apache-2.0, text CC BY 4.0. Only counts are published:...