Blog Posts

How to introduce new credentials, coordinate consumers, and retire old dependencies safely, with HMAC tokens as a detailed case study.

Learn why unique authentication tokens can still be easy to guess, with a simple Python example. Part 1 of a beginner's guide to backend authentication tokens.

Build a 37-character random token in Python. Learn how secrets.choice, string.ascii_letters, range, and join work together, one step at a time.

AI can write more code than ever. I'm thinking about the skills that help me decide what to build, catch what goes wrong, and own the result.

A real AI assistant does not just wait for commands. This post covers how OpenClaw runs scheduled jobs, responds to events, and completes background tasks without anyone opening a chat.

OpenClaw uses a plugin architecture to add AI models, messaging channels, and tools without touching the core. Nodes extend that reach to real devices like your phone and laptop.

Most agent tutorials jump straight to global installs. Here's the simpler path I actually use for internal tools: one Python app, one workspace, zero shared state.

When the workspace-local pattern isn't enough, you need a global skill. Here's how to write one with named tools that any agent can call.

How OpenClaw uses agents, session keys, and workspace scopes to run a personal assistant, a work assistant, and a briefing bot on one install without mixing their memory.

OpenClaw is an open-source AI assistant that runs across WhatsApp, Telegram, Slack, and your terminal as one coherent system. This post explains how the Gateway makes that possible.