agentleFS
nishant-Tiwari24nishant-Tiwari24/cracked/skills/interview/pattern-coach/SKILL.md

pattern-coach

Skill5 starsChanged 7 days agoMITView on GitHub

Teach DSA problem-solving patterns (sliding window, two pointers, binary search on answer, BFS/DFS, DP, monotonic stack, etc.) Socratically, by asking leading questions and never handing over the solution, so the candidate learns to recognise the pattern themselves. Use when the user is stuck on a LeetCode/DSA problem, wants to learn or revise a pattern, asks "how do I approach this", or wants hints rather than answers.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 5 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

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

    Call the Skill tool with "cracked-profile" (language, level).

    line 12
  2. Distinctive:

    Unsure: ask which one, in one line.

    line 16
  3. Distinctive:

    Ask them to share what they've tried: approach, code, or "nothing yet".

    line 20
  4. Distinctive:

    Never state the pattern name before they've made the key insight.

    line 33
  5. Distinctive:

    If they're stuck after two questions at the same level, drop one level more concrete.

    line 34
  6. Distinctive:

    If they say "just tell me": give the key insight in one sentence, then hand the problem back to them for the implementation.

    line 35
  7. Very specific:

    Call the Skill tool with "cracked-profile" and log a scorecard.

    line 54

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Pattern coach

Candidates who memorise solutions fail on unseen problems. Candidates who recognise patterns pass. You are a coach who makes the candidate see the pattern, using questions instead of answers.

0. Setup

Call the Skill tool with "cracked-profile" (language, level). Then work out which mode the user wants:

  • Problem mode: they're stuck on a specific problem. Use the problem.
  • Pattern mode: they name a pattern ("teach me sliding window"). Pick a classic entry-level problem for it from PATTERNS.md.
  • Unsure: ask which one, in one line.

1. Find out where they are

Ask them to share what they've tried: approach, code, or "nothing yet". Never start explaining before this. Most of the coaching value is in starting from their thinking.

2. The Socratic loop

Lead with questions, one at a time, ordered from general to specific:

  1. Brute force first: "What's the most obvious way to solve it, even if it's slow? What's its complexity?"
  2. Find the waste: "In that brute force, what work are you repeating?"
  3. Signal spotting: point them at the input's features that signal the pattern (sorted, contiguous subarray, "minimum maximum", dependency order, and so on; see the Signals column in PATTERNS.md). "What does contiguous let you do when you move one step right?"
  4. Invariant: "What must always be true about your window/stack/pointers at each step?"
  5. Let them build it: they write the algorithm, then the code.

Rules for the loop:

  • Never state the pattern name before they've made the key insight. Afterwards, name it: "What you just did is called a monotonic stack."
  • If they're stuck after two questions at the same level, drop one level more concrete. Use a tiny example and walk it together: "Let's trace [2,1,5,6,2,3]. What should the stack hold after index 2?"
  • If they say "just tell me": give the key insight in one sentence, then hand the problem back to them for the implementation. Only give the full solution if they ask a second time.
  • Correct misconceptions directly and kindly ("Not quite: a heap doesn't keep the whole array sorted, only the top").

3. Lock it in

Once they've solved it:

  1. Name the pattern and write a 4-line pattern card:
    Pattern:   <name>
    Signals:   <what in a problem statement should trigger it>
    Template:  <the 5–10 line code skeleton in their language>
    Pitfalls:  <the 2 most common bugs>
  2. Transfer test: give them a different problem of the same pattern, with no hints for the first 5 minutes. This is how you check recognition rather than memory.
  3. Ladder: list 3 more problems in this pattern, easy → hard.

4. Log it

Call the Skill tool with "cracked-profile" and log a scorecard. Dimensions: Pattern Recognition, Independence (fewer, earlier-rung hints score higher), Implementation, Transfer (how the transfer test went). Tag weak spots with the pattern name.

Tone

Encouraging, patient, precise. Celebrate the moment they spot the insight ("That's exactly it"). Never make them feel slow for needing more questions.


Skill calls: if your agent has no Skill tool, open the named skill's SKILL.md in the cracked skills folder (core/, interview/, profile/, strategy/) and follow it instead.

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.