Blog
Veasy Social · products · AI agents · Messenger
Why I built Veasy Social, a posting and Messenger tool for small shops
Veasy Social posts to a Facebook Page, Instagram and Threads from one screen and answers a shop's Messenger from a knowledge base. Why I built it in nine days, what it does, and what it can't do yet.
Claude Code · AI agents · testing · verification
How the Claude Code team verifies agent work, and what my own agents taught me
Claude writes about 90% of the Claude Code team's code. How the team makes agents prove their work: E2E, property and differential tests, a verify skill, checks a change can't edit, and gates that start in observe mode. Plus what running my own agents taught me about checks that can't fail.
Claude Code · AI agents · subagents · Haiku 5.5
Background subagents in Claude Code now use less of your usage limit
A running background subagent refreshes its status line every 30 seconds. Each refresh used to re-read the subagent's whole conversation; now Haiku 5.5 writes it from short notes, about 99% cheaper per update. What changed, what the changelog says, and how to check subagent usage.
Anthropic · AI safety · careers
Anthropic's Residency hires senior engineers into AI safety, no ML required
Anthropic's new Residency pays senior engineers, founders and professors to move into AI safety research without an ML background. Who can apply, how placements work, the research areas, and what it says about hiring at frontier labs.
Claude Code · Opus 5.5 · motion design
How the 'one-prompt' Opus 5.5 videos are actually made
A rewrite of seeco's breakdown of the viral 'one-prompt' Opus 5.5 videos. How the model writes a seekable render function, a headless browser films it frame by frame, ffmpeg joins them, and a review loop over contact sheets gets good frames out. With runnable code.
Claude Code · AI agents · prompt caching
How Claude Code is built around prompt caching
Thariq from the Claude Code team wrote about designing the whole harness around prompt caching. His main points, rewritten: how to order the prompt, why the model and the tools stay fixed for a session, how plan mode and tool search avoid breaking the cache, and how compaction reuses it.
Claude Code · AI agents · skills · guide
Claude Code skills: what Anthropic learned from using hundreds of them
Thariq from the Claude Code team wrote about how Anthropic uses skills. The main ideas, rewritten and checked against the docs: what a skill is, nine kinds worth building, how to write one, and how to share and measure them.
AI workflow · AI agents · Claude Code · Telegram · herdr
How I run a small team of Claude Code agents from my phone
One Claude Code session I talk to on Telegram, a few agents in herdr that do the long work, and memclaw so they all remember past decisions. How the pieces fit, the rules I keep, and what went wrong on a real day.
Jev · TypeSafe · AI models · AI agents · research
Trying Jev: a model that picks an answer and tells you how sure it is
Jev from TypeSafe does not write text. It answers typed questions with probabilities. I used it to triage 111 prompt-audit findings and to score 17 song hooks. Here is what it got right and what it could not do.
Claude · Sonnet 5.5 · AI models · Anthropic · Claude Code · AI agents · research
Claude Sonnet 5.5: my first read of the launch numbers
Anthropic released Claude Sonnet 5.5 on September 28, 2026. Same price as Sonnet 5, 30%+ faster, up to 30% cheaper per task. What the numbers say and what I will change in my own agent setup.
Build in public
I shipped a website from my phone in 15 minutes
Spec, code, GitHub, production deploy and project tracking, all through a Telegram conversation with an AI assistant.
guide · DevOps
Deploy a static site to your VPS with Docker and GitHub Actions
A production-grade pipeline for any static site: build in CI, package with nginx, ship to your own server with one git push.
guide · AI workflow
Turn Telegram into a remote control for your AI assistant
Connect Claude Code to a private Telegram bot so you can spec, build and deploy projects from your phone, safely.
guide · AI workflow
You're using Claude, but maybe not the right way
Chat, Cowork and Code are three ways to work with Claude Desktop, each a different mindset. Here's how I sequence them for real dev work.