How I Tested Hindsight Memory Without Fooling Myself
How I Tested Hindsight Memory Without Fooling Myself An agent with memory looks smarter in almost any...
Tutorials
“Hindsight Stopped My On-Call Agent From Repeating a Failed Fix”
"A practical incident-triage design that remembers failed fixes, not just successful ones, so an...
Per-agent tokens for MCP servers: what operators told me, and the design that came out of it
A few weeks ago I set out to give every AI agent its own cryptographic identity. Then I asked the...
Self-heal broken locators with a 2B model on your own GPU
Qwen3.5-2B with a LoRA heals broken UI-test locators in Optics Framework 1.11.3. A direct weight edit added 1.6 points. The LoRA added up to 7.
How a paid MCP server can charge AI agents safely (what we learned building one)
I'm the maker of AwesomeVideo, a hosted captioning API and MCP server that agents pay for with...
We let an AI red-team our hardware encryption. Here's what it found.
I build Crypt-in alone, in Bologna. It's file encryption anchored to a €5 ESP32-S3: the root secret...
Day-0 Model Support Is a Fork: What 26 NVIDIA Dynamo Snapshot Builds Say About Pinning Your LLM Serving Stack
I read 36 Dynamo release notes. Stable pins trail vLLM by a median 30 days, and 16 of 26 model snapshots sit on patched or custom engine bases.
Everyone Is Building AI to Replace Developers. Nobody Is Building the Plan for What Comes After.
Quick game. Open the launch post for any AI coding tool and count the verbs. Generates. Refactors....
SWE-bench Multilingual on Hugging Face: 15 of 34 Scores Are Not a Count Out of 300 Tasks
Audit of the 34-row SWE-bench Multilingual board on Hugging Face: 15 scores cannot be a count of 300 tasks, and the top rows sit inside noise.
When "Not Found" Doesn't Mean "Never Happened"
A lease expiry proves only that a worker stopped reporting. In five real-clock runs, a naive reconciler wrote the same ticket twice, and a late commit survived the lookup.
The New York Times v. Brave Puts AI Search Indexing Under Scrutiny
The New York Times Co. has filed a copyright complaint against Brave Software that places a growing...
How We Stitch Parallel AI Video Segments With Frame Accuracy
AI video models have a length limit, and the one we use renders at most 15 seconds per clip. That is...
Take a Claude Agent SDK agent from install to deploy with Agents CLI
How the Claude extension in Agents CLI takes a Claude Agent SDK agent from install to a validated Cloud Run deploy, with sessions, tracing and eval set up for you.
Ziff Davis Reports 37% Drop in Google Referrals as AI Visibility Reaches 2.7 Billion Impressions
Ziff Davis reported a sharp decline in organic Google referrals during the third quarter of 2026,...
Across Three Pull Requests to Three Different Projects, Only One Ever Got a Human Review
I sent three pull requests to three different open source projects over about three weeks this...
Building an orchestrator for Claude Code: recoverable before autonomous
For months, my AI-assisted development loop looked like this. I opened a chat and asked for a phased...
Your AI Coding Agent Saved Your Password to Disk. I Built the Tool That Finds It.
One click. "Yes, and don't ask again." Your agent runs the database command, and the password you...
Three weeks of watching MCP advisories appear and disappear
Three weeks ago I published a single-night scan of the top 500 MCP servers. Then I left the scanner...
I tested ChatGPT's PDF Content Credentials: what survives a re-save, a print and an edit
Ten ChatGPT-generated PDFs carried a C2PA manifest signed by OpenAI. I re-saved, re-rendered, rasterised and edited them to see what is left. Code included.
Why my Arabic debate-scoring engine refuses to give one overall score
In debate, two judges can score the same speech differently. Students get a number with no reasons,...
I tested whether eBPF helps diagnose slow AI agents. I wrote my predictions first and got many wrong.
A slow AI agent run is a chain of network calls: the model, tools, a database, outside APIs. An...
AI Visibility Is Real, but It Is Not a Standalone Performance Metric
AI visibility is measurable, but a growing body of research suggests that many dashboards overstate...
How Do You Stop Your AI Agent From Leaking Your Users' Data?
Your AI agent just summarized a customer's support ticket, drafted a reply, and queued it for review....
People are rebuilding whole languages in Rust with AI. I’m still asking why.
PHP, TypeScript, vibe coded and public. One ran up over $420k in API-priced tokens. Here’s what I...