Published: 2026-06-21
Analysis & perspective

Open Skills: Portable, Composable Agent Procedures Across Every Tool

Chapters / key moments (click to jump — plays here on the page)

Nate B Jones launches Open Skills — a public library of 31 reusable agent procedures (plus 7 runbooks) packaged as skill.md files that travel across Claude Code, Codex, Cursor, and any harness. The pitch: stop re-explaining how you work to every new agent — encode procedures once, scope them, compose them into runbooks, and bake verification into the contract.

Source video

"Your AI Skills Are Trapped | Here's How to Own Them" by Nate B Jones — Watch on YouTube →

Key Takeaways

Memory solves what your agent knows; 'procedural debt' is the second problem — the agent still doesn't know HOW you work (your voice, your testing standard, your definition of done).
A skill is a small skill.md folder with triggers (when to use / when NOT to), the job it owns, required tools, boundaries, expected output, and a verification standard — a reusable procedure, not a one-off prompt.
Skills are primitives; runbooks compose them into reliable workflows (e.g. voice memo → transcript → draft-in-your-voice → HTML page → published-and-verified).
Scope matters: personal skills (your voice, your publishing habits) live globally; project skills (test commands, safe selectors) live with the repo — so the files stop drifting between Cursor rules and CLAUDE.md.
Make verification part of the contract — 'do not call this done unless this evidence exists' — to turn agent output from review debt into real leverage.
A session-to-skill extractor closes the flywheel: at the end of a substantial session, capture any recurring non-obvious procedure as a new skill candidate instead of losing it to chat history.

What you can actually set up from this

Extracted from the video's own transcript — the specifics the original summary left out.

Reproducible steps

What a skill is
A small folder with a skill markdown file that says when to use it and when not to, what job it owns, which tools or files it uses, its boundaries, what the output looks like and how to verify it. A prompt is something you say once; a skill is something the agent can load again.
Compose skills into runbooks
Skills are primitives and runbooks are the chain. His voice-memo-to-published-page runbook chains media transcription, brain-dump processing, a personal voice skill, an HTML artifact builder and a site publisher. A release-day runbook chains current-information search, a release briefing, image generation, the publisher and a stakeholder update.
Choose the scope
Personal procedures (voice, publishing habits) live globally; project procedures (how to test this app, safe commands for this repo) live with the project. One markdown source of truth can feed Cursor rules, CLAUDE.md and Codex instructions instead of drifting copies.
Put verification in the skill
Define the proof before work starts: for browser QA, open the real route, check the console and mobile view, verify the changed workflow and capture screenshots; for research, compare sources and show dates. A session-to-skill extractor skill asks at the end of a session whether a recurring procedure is worth saving.

Gotchas

The launch described is a public library of 31 skills across seven categories plus seven runbooks, with a copy-paste setup prompt for each; we have not reviewed them.
Runbooks still need human judgment; skills aren't autonomy.
Most procedures should not become skills; the extractor is meant to answer no most of the time.