Aug 23, 2026
Publishing agent skills without leaking secrets
Skills are written next to .env files and real configs, which makes them a quiet channel for credential leaks. The failure modes, and what a safe publishing pipeline checks.
Blog
Agent skills, done properly
Aug 22, 2026
Your team's AI knowledge is dying in Slack
Prompts and agent instructions shared in chat have a half-life of a week. Why chat is where AI knowledge goes to die, and what an actual knowledge pipeline for agents looks like.
Aug 21, 2026
Skills vs. MCP servers vs. prompts: what goes where
Three ways to extend an AI coding agent — prompts, MCP servers, and agent skills — compared honestly, with a rule of thumb for choosing between them.
Aug 20, 2026
Writing SKILL.md: a practical guide
How to write a SKILL.md that agents actually trigger and follow — frontmatter that routes correctly, instructions that survive a fresh session, and the mistakes that quietly break skills.
Aug 19, 2026
Agent skills need versioning — here's why
Skills change agent behavior the way code changes production. Why unversioned skills are a silent reliability problem, and what pinning and rollback look like for SKILL.md packages.
Aug 18, 2026
How to share Claude Code skills with your team
Four ways teams distribute Claude Code skills — shared folders, git repos, copy-paste, and a private registry — and what each one costs you in practice.
Aug 17, 2026
What are agent skills? SKILL.md, explained
A plain-language explanation of agent skills — the SKILL.md packages that teach AI coding agents your workflows — and why they became the standard way to extend agents.