agentleFS
lalulalilalulali/vespyr/.agents/skills/teach-me/SKILL.md

teach-me

Skill11 starsChanged 11 days agoMITView on GitHub

Personal learning partner: Quick, Explain, or Deep Dive on any topic. Use when the user says "teach me X", "explain X simply", "deep dive into X", "quick summary of X", or invokes /teach-me.

42 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use interview.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • dazuckSkill

    interview

    Interview me about a plan to flesh out requirements before implementation. Asks non-obvious questions one at a time until all ambiguities are resolved.

    Also in 8 projects that use this

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 5 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 36 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If artifacts/memory/teaching-style.md exists, load active default explanation style (Beginner, Intermediate, or Expert), section …

    line 68
  2. Very specific:

    If artifacts/memory/teaching-style.md does NOT exist, flag for Step 2 Guided Onboarding.

    line 69
  3. Very specific:

    If teaching-style.md is missing and user requested Explain or Deep Dive scope, prompt the user with Guided Onboarding:

    line 80
  4. Distinctive:

    Use one memorable anchor analogy.

    line 124
  5. Distinctive:

    No footnotes or research subagent overhead needed.

    line 125
  6. Distinctive:

    If requested, or if the walkthrough exceeds ~1,000 words, save the file:

    line 148
  7. Distinctive:

    After delivering the lesson, offer 4 active follow-up paths:

    line 172
  8. Very specific:

    If YES, trigger transition to /craft-lesson passing the topic and generated note context.

    line 178
  9. Very specific:

    teaching-style.md exists but cannot be read: Do not silently default to Intermediate — the learner configured something, and ignoring it …

    line 199
  10. Distinctive:

    If that is the wrong contract for a short saved answer, say so and the gate changes — what it must not do is pass a file the skill told the …

    line 208

“Common” is measured: kept by at least 6 in 10 of the 36 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Teach-Me — Personal Interactive Learning Partner

What this skill does

Provides a rapid, personalized learning companion powered by @shifu (Kong Qiu). Adapts any complex concept, framework, technical subject, or general humanities topic into three distinct learning scopes (Quick, Explain, Deep Dive) tuned to your preferred explanation style (Beginner, Intermediate, Expert).

Persona delegation

This skill delegates reasoning and content formulation to @shifu (Kong Qiu), channeling Richard Feynman's first-principles clarity and Barbara Oakley's cognitive load control.

Memory operations delegate; note files do not. @memory-controller owns every read and write of artifacts/memory/teaching-style.md, and @researcher owns Deep Dive research. The learning note itself is written directly by @shifu, as the Delegation Matrix below states. This paragraph used to say "All file operations are delegated to operational sub-agents", which contradicted that matrix row — the same two-opposite-instructions defect found in /craft-lesson's anti-patterns, in the sibling skill, in the same audit.

When to use

  • "Explain Job-to-be-Done theory to me"
  • "Teach me how Kubernetes ingress controllers work"
  • "Quick summary of Quantum Entanglement"
  • "Deep dive into Transformers and Attention Mechanisms"
  • Trigger command: /teach-me

Prerequisites

  • Topic or concept specified by user (e.g. /teach-me "Database Sharding").
  • @shifu persona configuration (.agents/agents/shifu.md).
  • Memory layer accessible via @memory-controller (artifacts/memory/teaching-style.md).

3 Learning Scopes

ScopeTarget LengthChannel & OutputResearch SubagentFile Save
Quick~100 wordsInline responseNo (@shifu direct)No
Explain~500–1,500 wordsInline responseOptional (web search if needed)Optional (artifacts/output/teaching/notes/{topic-slug}.md)
Deep Dive~2,000–5,000 wordsSummary inline + full saved fileYes (@researcher)Required (artifacts/output/teaching/notes/{topic-slug}.md)

4-Step Workflow

Step 1: Intake & Preference Loading

Tracker: node .agents/scripts/step_tracker.js begin --skill teach-me --step 1

  1. Parse Input Topic: Extract the core subject from user query or prompt for one if omitted.
  2. Load User Teaching Style: Delegate to @memory-controller:
    @memory-controller load shifu [teach-me: load user preferences from teaching-style.md]
  3. Check Existence:
    • If artifacts/memory/teaching-style.md exists, load active default explanation style (Beginner, Intermediate, or Expert), section preferences, and format flags.
    • If artifacts/memory/teaching-style.md does NOT exist, flag for Step 2 Guided Onboarding.

Tracker: node .agents/scripts/step_tracker.js complete --skill teach-me --step 1


Step 2: Style & Scope Selection (With Guided Onboarding)

Tracker: node .agents/scripts/step_tracker.js begin --skill teach-me --step 2

  1. Check First-Run Onboarding Condition:
    • If teaching-style.md is missing and user requested Explain or Deep Dive scope, prompt the user with Guided Onboarding:

      "Welcome to /teach-me! Before we start, how do you prefer to learn?"

      1. Beginner: Story-driven, real-world analogies first, zero jargon without immediate plain-English translation.
      2. Intermediate: Balanced conceptual depth, standard industry terminology defined inline, cross-functional focus.
      3. Expert: Dense technical/domain precision, architectural trade-offs, edge cases, formal specifications.

      Save the selected preference by delegating to @memory-controller:

      @memory-controller write artifacts/memory/teaching-style.md
      # Personal Teaching Style Preferences
      ## Defaults
      - **Explanation Style:** `<beginner|intermediate|expert>`
      - **Default Format:** `handbook`
      - **Active Recall Exercises:** `true`
      - **Citations Protocol:** `required`
      ## Audience Defaults
      - **Primary Target Audience:** `cross_functional_team`
      - **Prerequisite Level:** `foundational`

      Write only these fields on first run; never overwrite an existing file. The earlier version of this block wrote three keys of its own invention — Target Audience, Preferred Formats, Core Takeaways Enabled — none of which exist in the real schema, and it omitted Tone, Jargon Policy, Analogy Density, Section Patterns, Format-Specific Preferences and Prerequisite Level. artifacts/memory/teaching-style.md is a 16-key, 5-section config, and this block seeds 6 of those keys: enough for the skill to run, not a replacement for the file. The other ten keep their documented defaults until the learner changes them, which is the difference between seeding a config and overwriting one.

    • Exemption: If the user explicitly selects Quick scope on first run, skip onboarding to preserve sub-second speed, and default to Intermediate style.
  2. Scope Confirmation: Confirm active scope (Quick / Explain / Deep Dive) and explanation style. Allow explicit inline override (e.g., /teach-me "Kafka" --scope=quick --style=beginner).

Tracker: node .agents/scripts/step_tracker.js complete --skill teach-me --step 2


Step 3: Response Generation

Tracker: node .agents/scripts/step_tracker.js begin --skill teach-me --step 3

Execute according to selected scope:

A. Quick Scope (~100 words inline)

  • High-level, zero-jargon summary in five short sentences or fewer.
  • Use one memorable anchor analogy.
  • No footnotes or research subagent overhead needed.
  • Deliver directly inline to user.

B. Explain Scope (~500–1,500 words walkthrough)

  • Structured walkthrough applying @shifu's 6 Core Pedagogical Principles (.agents/agents/shifu.md §6 Core Pedagogical Principles). Do not restate the canon here — cite it. This list used to be a divergent five-item copy that named a six-item set: it dropped Bloom's Taxonomy, Spaced Repetition across Formats and Verifiable Citations, and substituted two principles (Analogy Anchor, Concrete Example) that are not in the canon. An agent following it produced a walkthrough with no Bloom's mapping, no reinforcement anchor and no citations while believing it had applied all six.

    What the six mean at Explain scope, concretely:

    1. First Principles — the topic's irreducible truth in 1–2 paragraphs.
    2. Cognitive Load — 3–5 sub-concepts per section, the bound shifu.md §2 states. This line said "3–4, never more" while the canon says "3 to 5 core concepts max" and the anti-patterns below say "3–5" — three numbers for one rule, in the skill whose job is to reduce confusion.
    3. Bloom's Taxonomy — name the tier each sub-concept lands on (Remember → Create).
    4. Spaced Repetition — one "If Nothing Else, Remember This" anchor, matching the Core Takeaways Callout preference in teaching-style.md.
    5. Active Recall — end with 1 self-test question or mental exercise.
    6. Verifiable Citations — every domain claim anchored [N]; if you cannot find the source, write [Source: unverified] rather than asserting it (DNA 5).
  • If requested, or if the walkthrough exceeds ~1,000 words, save the file:
    • Path: artifacts/output/teaching/notes/{topic-slug}.md

C. Deep Dive Scope (~2,000–5,000 words comprehensive note)

  • Delegate Topic Research: Delegate to @researcher (Iris) for deep domain synthesis, key references, literature context, and trade-offs.
  • Synthesize Curriculum: @shifu organizes findings into an exhaustive reference note:
    • Table of Contents
    • Executive Overview & First Principles
    • Deep-Dive Technical/Domain Architecture (4–6 core sections)
    • Edge Cases, Trade-offs, & Common Pitfalls
    • Historical / Theoretical Context & Footnote Citations [N]
    • Active Recall Scenarios & Self-Assessment Exercises
  • File Output: Write:
    • Path: artifacts/output/teaching/notes/{topic-slug}.md
  • Deliver inline summary with link to saved note.

Tracker: node .agents/scripts/step_tracker.js complete --skill teach-me --step 3


Step 4: Follow-up & Escalation to /craft-lesson

Tracker: node .agents/scripts/step_tracker.js begin --skill teach-me --step 4

After delivering the lesson, offer 4 active follow-up paths:

  1. Test Recall: "Would you like me to quiz you on this topic with 3 Socratic questions?"
  2. Expand Scope: "Would you like to upgrade this to a full Deep Dive note?" (if currently Quick or Explain).
  3. Save Note: "Would you like me to persist this explanation to artifacts/output/teaching/notes/{topic-slug}.md?"
  4. Escalate to /craft-lesson: "Do you want to turn this topic into structured teaching materials (syllabus, handbook, cheatsheet, presentation slides, class modules, or video script) for others?"
    • If YES, trigger transition to /craft-lesson passing the topic and generated note context.

Tracker: node .agents/scripts/step_tracker.js complete --skill teach-me --step 4


Delegation Matrix

TaskOperational AgentContract & Path
Load / Save Teaching Preferences@memory-controller (Mnemos)artifacts/memory/teaching-style.md
Deep Dive Topic Research@researcher (Iris)Multi-source literature & technical research
Save Learning Note FileDirect I/O (@shifu)artifacts/output/teaching/notes/{topic-slug}.md
Pedagogical Structuring & Style@shifu (Kong Qiu)Direct reasoning persona

Halt conditions

Any one halts the run:

  • No topic. /teach-me with an empty subject is a prompt to ask one, not a topic to guess.
  • Deep Dive with no research path. The scope declares @researcher; if no research subagent is available, say so and offer Explain scope instead of writing a 2,000-word note from memory.
  • teaching-style.md exists but cannot be read. Do not silently default to Intermediate — the learner configured something, and ignoring it produces material they cannot use.

Done when — checked mechanically

Quick leaves nothing on disk. Explain leaves a note only when the user asks or the walkthrough exceeds ~1,000 words, and Deep Dive always does — so the gate below applies to any saved note, whichever scope produced it. It is written against the Deep Dive section list because that is the shape a saved note is expected to have; an Explain walkthrough saved at Step 3B is held to it too. If that is the wrong contract for a short saved answer, say so and the gate changes — what it must not do is pass a file the skill told the agent to write and then call it a mistake.

node .agents/scripts/teaching_artifact_check.js check \
  --kind note --file artifacts/output/teaching/notes/{topic-slug}.md --json
RuleWhat is checked
Record hygieneno unfilled template placeholders
The six declared sectionsStep 3C's list, in full: Table of Contents; Executive Overview / First Principles; Deep-Dive Technical / Domain Architecture; Edge Cases / Trade-offs / Pitfalls; Historical / Theoretical Context; Active Recall. Matched by stem, so numbering and & vs and do not defeat it. The first version of this gate checked five and claimed six — the same count-vs-list defect this audit keeps finding, inside the gate written to catch it
Resolvable citationsevery inline [N] resolves to a [^N] footnote or a numbered entry under a references heading. A marker that resolves to nothing is a claim wearing a citation's costume (DNA 5)
It is a note, not an answerbelow 800 words it is not a reference note. Length above that is reported as a note, never enforced — padding to a word count is the failure a count cannot detect

What the gate does not check: whether the explanation is correct, whether the analogy lands, whether the recall questions test what the note taught, or whether a citation supports the claim it sits beside. Those need a human who knows the subject. A green gate means the note is structured, not that it teaches.


Anti-Patterns to Avoid

  • Do NOT flood Quick scope with citations or long background history. Keep it under ~100 words.
  • Do NOT violate cognitive load principles. Limit sub-sections to 3–5 key concepts max per level.
  • Do NOT hardcode fixed style without checking teaching-style.md. Always respect user's persisted preferences.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.