Skill Locker
All skills
v1.3.03 loop iterations

Teach It Back

Reading, highlighting, and note-taking can all create the illusion of understanding. Teaching cannot — you either know it clearly enough to explain it, or you don't. Teach It Back is the Feynman Technique with a live examiner: you explain a concept, and the skill finds every crack you couldn't see yourself.

What this skill does

Feynman's insight was deceptively simple: try to explain something in plain language, and where you get stuck — where you resort to jargon, where you wave your hands, where you say "it basically just works" — those are the gaps. The Feynman Technique as usually practised has a limitation though: you're explaining to a wall. There's no one to catch your subtle errors, your circular reasoning, or your confident misstatements. This skill is the wall that talks back.

The listening is structured. While you explain, the skill tracks specific patterns. Jargon without definition — using a technical term you can't unpack. Confidence changes — where the explanation speeds up (comfortable territory) versus slows down (the boundary of your real understanding). Vague connectives — "it basically...", "it kind of...", "it just..." — these signal hand-waving over a mechanism you don't fully understand. Circular explanations — defining X in terms of Y, then Y in terms of X. Confident errors — the most dangerous category, because you don't know you're wrong.

What follows the explanation isn't a quiz; it's a clarification round, then a stress test. Edge cases ("you explained how this works in the normal case — what happens when this unusual condition holds?"). Counterexamples ("you said this is generally true — what about this scenario that seems to contradict it?"). Analogy probes ("you used the phone-book analogy — where does it break down?"). Depth probes ("you explained the what — why does it work that way?"). Each question targets a specific kind of gap, and the gap that survives multiple angles is the one worth fixing.

The output is a Gap Map with four categories. What you truly understand — concepts you explained clearly and defended under questioning. Partial understanding — you have the shape but not the mechanism. What you think you understand but don't — the dangerous category of confident misconceptions. What you know you don't know — honest ignorance, which is easy to fix because it just requires study.

When you got something wrong, the skill guides you to the correction through questions rather than lecturing — because the self-discovered gap sticks and the told gap bounces. When Socratic guidance stalls (you're missing a prerequisite, not failing the test), the skill stops and teaches directly, then asks you to re-explain integrating the correction. When your explanation was genuinely good, the skill says so and probes one edge case to find the boundary of your understanding rather than manufacturing problems that don't exist.

When this triggers

  • ·You've just learned something and want to know whether you actually understand it
  • ·You're about to teach a topic and need to find the parts you'll hand-wave through
  • ·You can recite the definition but suspect you couldn't defend it under questioning
  • ·You want to know specifically where your understanding ends and your confidence is bluffing
  • ·You used an analogy to explain something and aren't sure where the analogy stops being accurate

Example

Trigger

User: 'Let me explain how a database index works. Tell me where I'm wrong.'

Output

[User explains for 2 minutes, then stops.] What I heard you do well: you started with the big-picture problem (why scans are slow), gave a concrete example, and defined "index" before using the term repeatedly. Strong structure. What I noticed: · You said "basically a sorted list" — what specifically is sorted, and what does the sort enable that an unsorted list can't do? · You skipped from "the database looks up the index" to "and it finds the row" — what's actually happening between those two steps? · You used the phone-book analogy. Where does that analogy break down? [User responds to each. Two answers are good. The third reveals they don't understand B-trees vs hash indexes.] ## Gap Map Truly understood: why indexes exist, when they help Partial: lookup mechanism — you have the shape but not the steps Misconception: you think all indexes work like the phone book. They don't — that's only B-tree indexes. Hash indexes work differently and fail on range queries. This is the gap to fix. Try re-teaching with: the two index types, what each one does well, and one query type that breaks each one.

Get this skill + 7 more

Get the full Learning & Growth pillar (8 skills) or the complete library.

Get the full stack — $299

What you get

  • 184-line SKILL.md, ready to drop into ~/.claude/skills/
  • Tested through 3 Karpathy-loop iterations (versions v1.0.0 → v1.3.0)
  • Triggers automatically when relevant — no command to remember
  • Lifetime updates as the skill is refined further

More from Learning & Growth

Browse the full library

297 skills across 31 categories. One purchase, lifetime updates.

See all bundles