Trying Agent Autoloop: Events, Decisions, and CLI Agents
A hands-on look at Agent Autoloop, with a local two-turn demo, actual CLI evidence, and the limits of what I tested.
Recent posts
Do Agent Skills Actually Help? I Tried Writing One
What a SKILL.md adds, how to write and install one, and what a small before-and-after test actually showed.
Does AI Help Children Learn, or Just Finish Their Homework?
What studies say about AI, independent thinking, and cheating—and a practical routine for keeping the learner involved.
Using AI for Investing and Trading: Where I Would Start
A practical workflow for AI-assisted investment research, strategy testing, and paper trading, with the benefits, failure modes, and one misleading trading s...
Trying CloakBrowser: Familiar Playwright Code, a Different Chromium
What CloakBrowser changes, when I would use it, and a small Python demo I ran with a real screenshot.
REA: Following a Feature Through an App with an AI Agent
A practical look at REA, the difference between its skill and MCP tools, and a small Electron example you can inspect without installing Electron.
Reading a Zalo Desktop Backup: From .zl.zip to Searchable Messages
Why a Zalo export would not unzip, how I verified its AES-CBC format, and how I recovered a readable archive. Includes a synthetic example.
DeepSeek Harness: What ‘Everything Is a Plugin’ Actually Means
I went through the new DeepSeek Harness repo and documentation because the tagline sounded almost too broad:
Paperclip: Managing AI Agents Like a Team
I read through paperclipai/paperclip because the idea sounded ambitious: an app for managing AI agents at work.
NocoDB: A Fast Way to Put a Database Online
I read through nocodb/nocodb because I wanted to understand the claim:
Headless CMS vs Building the Backend Yourself
I used to put “headless CMS” and “backend API” in the same bucket.