axel-pm/watch-n-learn / skills/watch-and-learn
watch-and-learn
VerifiedWatch a how-to / tutorial / demo video and turn the procedure it teaches into one or more working Claude Code skills. Use when the user wants to "turn this video into a skill", "watch this tutorial and make a skill", "learn how to do X from this video", "build a skill from this screencast/demo/talk", or capture a repeatable workflow shown in a video. Sends an agent to watch the video (local file or URL) with a local, free, no-API toolchain, breaks the procedure down, authors the skill(s), and verifies they actually work.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 7.3 KB |
| references/procedure-spec-template.md | markdown | 1.5 KB |
| references/qualifying.md | markdown | 2.9 KB |
| references/skill-authoring.md | markdown | 2.5 KB |
| references/validation-checklist.md | markdown | 1.8 KB |
| references/watching.md | markdown | 4.6 KB |
| scripts/install-tools.shexecutable | script | 1.7 KB |
| scripts/validate_skill.py | script | 4.4 KB |
| scripts/watch.py | script | 16.3 KB |
references/qualifying.md
View on GitHub# qualifying.md — is this video worth a skill?
The gate before any real work (step 1 of the skill). Authoring a skill costs effort and adds a file
that loads into every future session; a weak skill is worse than none. Spend thirty seconds here first.
Do this **cheaply**: run `watch.py --transcript-only` (talks/narrated videos) or just read the title +
description, and judge against the signals below. You don't need the full frame-by-frame watch to decide.
## PASS — worth authoring
A skill is worth it when the video teaches a **concrete, repeatable, transferable procedure** — the kind
of thing you'd want an agent to just *do* later without re-deriving it. Signals:
- It shows a **specific how-to**: setup, a workflow, a build, a fix, a configuration — with actual steps.
- The steps **generalise** beyond this one video (a real user could apply them to their own case).
- You can imagine **triggering it later** ("do the X setup", "run the Y workflow") and being glad it exists.
- The procedure is **non-obvious or easy to get wrong** — capturing it saves real time or mistakes.
## FAIL — don't author (say why, suggest a better home)
Stop and report instead of forcing a skill when the video is:
- **Opinion, commentary, or entertainment** — a talk that inspires but teaches no procedure. → Better as
notes or a summary, not a skill.
- **A one-off** with nothing reusable — a demo of a specific dataset/account/repo that won't transfer.
- **Too trivial** — the "procedure" is a single obvious command or a menu click. A skill adds ceremony
for no gain.
- **Already covered** — you (or the user) already have a skill for this. → Point at the existing one.
- **Not actually a procedure** — a product announcement, a keynote, a vlog. → A reference doc at most.
## When it fails
Don't quietly build a mediocre skill. Say plainly: *"This video is <reason>; it isn't a good skill
candidate. Here's what I'd do instead: <notes / reference doc / point at existing skill>."* Then stop.
## Escape hatch
If the video is **obviously** a strong how-to (a clear tutorial with clean steps), don't belabour the
gate — note "clearly a good candidate" in one line and move on to the watch. The gate is there to catch
the bad cases cheaply, not to add ceremony to the good ones.
## Worked examples
- *"Setting up Postgres row-level security with policies"* screencast → **PASS**: concrete, repeatable,
transferable; a great `configure-postgres-rls` skill.
- *"My thoughts on the future of AI agents"* conference keynote → **FAIL** (opinion): capture 3 bullets
of notes if useful; no skill.
- *"Deploying MY portfolio site to MY Vercel account"* walkthrough → **FAIL** (one-off): the reusable
bits are just "vercel deploy" — too trivial and too specific.
- *"Debugging a flaky test with git bisect and reproduction harnesses"* talk → **PASS**: the *method*
transfers even though the bug doesn't; a solid `bisect-flaky-tests` skill.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |