A newsletter for practitioners building with AI agents. Real implementations, actual code, honest assessments of what’s working and what isn’t. Three editions per week, under 5 minutes each.
- Monday Build — Deep dives into building agents, with real code and walkthroughs
- Wednesday Stack — Tools, frameworks, and infrastructure tested in production
- Friday Signal — What happened this week that changes how you build agents
Every edition includes Quick Hits (3 notable developments) and One Thing to Try (a single actionable step for your week).
The Agent Stack #003 — Friday Signal
OpenAI acqui-hired the creator of OpenClaw this week. 140,000+ GitHub stars, the most popular open-source agent framework, absorbed into the company that’s racing to dominate the agent layer. Here’s what it means for anyone building agents right now.
What Actually Happened Peter Steinberger, creator of OpenClaw, is joining OpenAI. The project had 140K+ stars on GitHub and had become the default starting point for developers building multi-agent systems. Sam Altman announced it directly.
...
The Agent Stack #002 — Wednesday Stack
I took 10 popular MCP servers, configured them exactly as their official READMEs tell you to, and ran a security scan. The score: 2 out of 100. Not because I misconfigured anything. Because the docs tell you to do things like this:
{ "env": { "GITHUB_PERSONAL_ACCESS_TOKEN": "ghp_your_token_here" } } Your GitHub PAT, sitting in a plain text JSON file that any application on your machine can read.
...
The Agent Stack #001 — Monday Build
Last week I asked an AI to scan my MCP server configs for security vulnerabilities, build a CLI tool to automate the checks, add baseline diffing, and publish it to npm. It did all of it across a few sessions. No copy-pasting from Stack Overflow. No scaffolding by hand. Conversations that produce working software.
That’s not a demo. That’s my normal workflow now.
...