Последние новости

Five ways an x402 paid endpoint can serve content it wasn't paid for, and a CI check for each

x402 turns HTTP 402 into a real payment flow: your server answers 402 with payment requirements, the client pays, and a facilitator verifies and set…

The agent "fixed" the race condition with a 200ms sleep. CI went green. Tuesday morning it fell over.

The flaky test was the kind that everyone on the team had learned to rerun. it ( ' replays the event into the read model ' , async () => { await…

The agent wrote tests that mocked the function under test — and CI went green

The bug report came in on a Tuesday morning: a user's avatar URL was being rendered as undefined in the profile dropdown. The code that builds the p…

I Let an LLM Agent Search for Trading Strategies. A Simple Heuristic Beat It.

I gave DeepSeek V3 forty chances to find a profitable trading strategy. A simple heuristic found +52% net. The LLM's early proposals lost money. Aft…

It sent gemini-2.5-pro and reported gemini-2.5-flash

A response's model field is not decoration. It is what your logs record, what your cost attribution divides by, what a router uses to pick the next…

Game Optimization Methodology, or How 3 Days of Building a Test Harness Gave Me More FPS Than a Month of Writing Code

About me I'm a mobile game developer. Most of my work is gameplay programming and building tools for development teams. I like Lua, C++, Godot, and…

The drift a scanner can't see: when the instruction is stale, not hostile

Dhanush N's agentcheck targets a real thing: instructions that arrive carrying an attack. A tag-character payload, a poisoned rule, a server that re…

I fuzzed 50,000 dollar amounts across my Python JS ports and found a silent 1.2% rounding bug

I maintain ~10 small zero-dependency CLI tools for freelancer admin (invoicing, late fees, currency conversion, tax estimates) — some in Python, som…

Your AI agent says "done". Your tests say 3 failing. The Stop hook that settles it

You ask an agent to fix a bug. It edits four files, writes a cheerful summary, and says the task is complete. You run the tests and three of them fa…

Android Version Code Strategy: versionCode vs versionName Done Right

Every release you ship to Google Play carries two version identifiers, and mixing them up is one of the most common — and most avoidable — release m…

Jepsen: Orkes Conductor 5.3.2

What checking 244 calculators against their sources actually found

I run CalculateEveryday , a free site with 244 calculators: paycheck, mortgage, BMI, unit conversion, chemistry, statistics and so on. Each page sho…

Census the Names a Function Reads Before You Extract It

Most extractions do not fail in review. They fail two weeks later, when a test hits a module global nobody remembered. The repair costs more than th…

Email Automation Needs a Dry-Run Contract

Email automation often fails before the email provider is actually the problem. A test creates a user, requests a verification message, waits for a…

Your Code Ran on the Free Box. That's Not an Environment Match.

Your tests pass on your laptop. The same commit then dies on a plain Ubuntu box with error while loading shared libraries: GLIBC_2.34 not found . Wa…

48-Hour Field Notes: My Job Had 30 Seconds. I Gave the Model 8 and Kept the Rest.

My batch job had a 30-second deadline, and the model call happily ate all 30 of them. The scheduler killed the worker mid-flight, the log ended on a…

A Frozen Test Can't Justify a Patch: Lease-Bound Freezes and Witnessed Properties for Agent Patches

The most expensive test in an agent patch is the one nobody runs. It sits in the suite, green, on a quarantine list, and the patch quietly points at…

Freeze the Export Surface Before a Messy Refactor: A Manifest Gate in Python

The smallest safe change is a measurable claim. Most refactors fail because nobody defined "small." A rename here. A default flipped there. Six week…

Fixing a 10-year-old fade race in jQuery Modal (#120), and a drag-select bug nobody filed until 2021

I keep one habit: find a bug in an abandoned open-source project, prove it with a test harness, ship the smallest fix that makes the harness pass. H…

The Model May Draft the Prose. The Repo Must Confirm the Facts.

A doc said the retry limit defaults to 5. The code said 3. The env var in the doc had been renamed two releases earlier, so anyone who copy-pasted t…

Shuffle the Golden Cases: Catching Eval State Leaks That Order Hides

Shuffle the Golden Cases: Catching Eval State Leaks That Order Hides A patch-eval suite can be green for months and still be lying. The lie usually…

A Model Fallback PR: What to Trust, What to Revert, What to Test

A fallback PR lands at 09:12 on a Monday. CI is green and the description reads "adds model fallback for reliability." The diff spans 340 lines acro…

How to Test AI Agent Crash Recovery Locally with Dapr Workflow

Introduction Most advice on testing AI agents is about the answers, such as evals and golden datasets. Crash recovery needs its own test. If your ag…

AI agent observability: what to record and why

AI agent observability is the practice of recording what an agent tried, which evidence shaped each step, and what the user experienced, so you can…

Your Free Server Forgets Everything: Six Deploy Myths and a Restart Drill

You ship a small demo on a free server. It works. You go to sleep. Next morning the app has no data. Nothing crashed. Nothing paged you. The process…

The Patch Added a Sleep: Reviewing Timing and Retry Hunks Before Merge

The Patch Added a Sleep: Reviewing Timing and Retry Hunks Before Merge Conclusion first: most arguments about "flaky tests" start after merge. The c…

Characterize the Boundary, Not the Module: An Import-Graph Gate for Messy Refactors

Characterize the Boundary, Not the Module: An Import-Graph Gate for Messy Refactors The rule: characterize only the symbols other modules import. Th…

How to Publish a Unity Game on Google Play: Build Settings, Keystores, and AAB Export

Unity developers live in the editor, but the path to publish a Unity game on Google Play runs through Android build tooling that Unity mostly hides…

A benchmark shipped 2,700 agent trajectories. I recomputed its leaderboard.

Most eval papers give you a table and a PDF. LSCBench gives you the runs. LSCBench is a secure-code benchmark out of NUS and the Qwen team: 150 real…

48-Hour Field Notes: Every Restart Restarted My Batch Job, So I Added Leases

Two days into a batch classification run, I opened the log and saw line one again. Not a continuation of line one — the same line one, with a new ti…