- Home
- Skills
- Education & Learning
- code coach
More screenshots
Works with the AI Agents you already use
code coach
Ask any AI "fix my exercise" and it does — in 4 seconds.
$5
code coach
Example session with this skill installed
I'm learning Python in my intro course and my exercise doesn't work — can you fix it? It's supposed to add up all the scores in a list but it prints 0 every time. I don't want to just get the answer if that makes me a worse programmer... my professor checks that we can explain our code. But I'm stuck and frustrated.
scores = [72, 85, 91, 68]
total = 0
for s in scores:
total == total + s
print(total)
- Read your context and instructions
- Compiled the code coach
Summary
Ran a full coaching session on the broken accumulator exercise: calibrated in 2 probes, guided up the hint ladder —
the coach never wrote the fix; the learner typed it — verified with a transfer exercise and a teach-back. The real bug was a misconception, not a typo.
Calibration (2 probes, before touching the bug)
- Probe 1 (flow): learner correctly explained
total = 0runs once, before the loop → advanced beginner, tracing is solid - Probe 2 (model): learner couldn't distinguish
=from==→ misconception located: assignment vs equality - Mode agreed: coach mode — hints only, learner drives
Hint Ladder (what actually happened)
| Rung | Coach said | Result |
|---|---|---|
| 1 — Nudge | "Read each character of the line inside the loop — one of them does a different job than you think." | Stared, couldn't see it |
| 2 — Question | "What does == ask, versus what = does?" | 💡 "So total == total + s asks a question and throws away the answer?" |
| 3 — Fragment | Same accumulate-pattern on a 4-line counting example (different domain) | Learner mapped it and fixed their own line: total = total + s → prints 316 ✓ |
Practice & Verification
- Transfer rep: predict-the-output on a similar
==trap with strings — learner got it cold ✓ - Teach-back: "why did
==give you 0 every time?" — learner explained in own words, no parroting ✓ - Professor-proof: the learner can now explain every line, because they wrote every line
Session Log
advanced beginner · Python · course goal · coach mode
Mastered:=vs==(explained back + transferred)
Misconception fixed: "assignment and equality are interchangeable"
Next: word-frequency counter drill — same accumulator thinking, +1 difficulty
Next Steps
- Bring the next exercise — calibration carries over within the topic
- Say "quiz me" anytime for a predict-and-run warmup
- If a deadline hits, switch to answer mode with one sentence — no guilt, takeaway included
Connects securely to your tools. The creator never sees your data.
About this skill
Your AI writes great code. And teaches nothing. Ask any AI "fix my exercise" and it does — in 4 seconds. You paste, it works, you learn nothing. Two weeks later you're stuck on the same class of bug, and the honest name for what's happening is answer dependence: the vending machine is fast, but you're not getting smarter. Tutorials have the opposite problem — they're generic. They don't know that you keep confusing = with ==, that you've never truly traced a loop, that your real gap is one model, not a whole course. What this skill does differently This skill turns the agent into a coach with a method, borrowed from the best teaching traditions and 500k★-level open curricula:
- Calibrate — two quick probes (or reading your code) place you on a level ladder and find the actual gap. No assumptions from "I'm a beginner".
- Contract — pick your mode: coach (hints only, you drive), hybrid (coach explains, you apply), or answer (you opt out; adults are allowed). Switchable anytime.
- Hint Ladder — stuck? You get a nudge, then a question, then the pattern on a different tiny example — one rung at a time. The coach never writes your code.
- Practice Design — exercises chosen to hit your diagnosed gap (predict-and-run, debug hunt, build-small, refactor), difficulty held in the zone where you fail ~30% and grow.
- Teach-Back — a concept counts as learned only when you can explain it in your own words or apply it cold. Parroting doesn't pass.
- Session Log — a 10-line bookmark: what's mastered, what's shaky, one exercise for next time. Who it's for
- CS students & bootcampers who want to pass their own exams, not submit AI homework
- Interview preppers who need to think out loud, not recognize answers
- Self-taught developers tired of tutorial hell and copy-paste growth
- Parents & mentors using AI to tutor without doing the homework for the kid The honest part Sometimes you just need the answer — deadline, real job, whatever. This skill doesn't guilt-trip you: answer mode is one sentence away, and the coach still leaves you a 60-second takeaway. Teaching has a wrong time too. The method exists for when learning IS the job.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 8 days ago
Permissions
File Scopes