____ ____ ____ ____ __ __ ____ ( _ \( _ \( __)( _ \( \/ )( __) ) __/ ) / ) _) ) __/ ) ( ) _) (__) (__\_)(____)(__) (_/\/\_)(____) ┌────────────────────────────────────────┐ │ $ prepme -- CV + JD loaded │ │ > your questions ready │ │ > cracking the interview ........ │ └────────────────────────────────────────┘
Feed it your résumé and the job description. Walk into the interview already knowing the questions.
prepme is an agent skill — drop it into any AI agent that supports skills — that turns a CV + JD into a focused, interactive study sheet of the questions you're most likely to be asked, then hands you a one-click AI deep-dive for each one.
It's the interview equivalent of getting the exam paper the night before, except entirely legal and you still have to do the studying.
No generic "tell me about yourself" filler. prepme predicts what this interviewer would actually ask this candidate, and it does it in two deliberate halves:
- The JD half — the knowledge they're hiring for. It strips the company's product fluff off each requirement and asks about the transferable fundamentals underneath. A "real-time payments platform on Kafka" JD becomes a clean question about Kafka delivery semantics — not "design our payments pipeline," because nobody learns event ordering from a job ad.
- The CV half — the work you actually did. It mines your projects and the tech you listed, then politely calls your bluff: what you built, why, the tradeoffs, and whether the word "expert" on line 12 of your résumé can survive three follow-up questions.
Every question is tagged Foundational / Core / Advanced, justified by why it gets asked, and pre-loaded with the 2–4 follow-ups an interviewer reaches for the moment they smell hand-waving.
A single self-contained HTML file — no server, no CDN, no "please wait while we load." Open it in any browser and:
- Tree-structured layout — questions branch into their follow-ups, so the "and then they'll ask…" chain is obvious at a glance.
- Click to copy — one click drops a ready-to-paste, language-matched deep-dive prompt onto your clipboard (best answer, bonus points, follow-up handling, ASCII diagrams) for any chat AI.
- Answer-driven progress — a card only flips to done once you've actually saved an answer for it (via
anslog, below). The progress bar counts answered questions, so it reflects real understanding — not how many buttons you clicked. - Filters — by category, by difficulty, or hide everything you've already conquered.
- Speaks your language — output language is auto-detected from your documents. Interview conducted in 中文? You get 中文, with the tech terms left mercifully untranslated.
In your AI agent, point it at your two files:
prepme — CV: ~/Documents/resume.pdf JD: ~/jobs/acme-backend.txt
…or just ask in plain English: "prepare interview questions from this resume and job description."
Out comes interview-prep.html. Open it. Start drilling. Try not to peek at the follow-ups first.
Once you've worked a question through with an AI agent and you're happy with the answer, ask your agent to log it:
log this answer (or: "save this answer to my study sheet")
That fires the companion anslog skill, which writes a clean, self-contained answer page under
answers/ and links it back into your study sheet — the card flips to an answered state with a
View answer button on reload. anslog only ever writes inside answers/; it never touches
interview-prep.html, so links survive even if you regenerate the sheet. prepme writes the
questions; anslog keeps the answers.
Two skills, one feedback loop. prepme writes the questions; you drill each one with an AI; anslog files the answer back into the sheet — which is exactly what flips the card to done.
you ──"prepme — CV + JD"──▶ ┌──────────┐
│ prepme │ designs the questions
└────┬─────┘
▼
interview-prep.html
(your study sheet)
│
┌──────────────────────────┼──────────────────────────┐
▼ ▼ ▼
[ Q: Kafka ] [ Q: your billing ] ...more questions
[ delivery sem... ] [ service rewrite ]
[ [ ] not answered ] [ [ ] not answered ]
│
│ 1. click [Copy prompt] ──▶ deep-dive prompt on clipboard
▼
┌───────────────┐ 2. paste into any chat AI, discuss,
│ AI agent │ sharpen the answer until you're happy
└───────┬───────┘
│ 3. "log this answer"
▼
┌──────────┐ writes answers/<id>.html + updates answers/answers.js
│ anslog │ (never edits interview-prep.html)
└────┬─────┘
│ 4. reload the sheet
▼
[ Q: Kafka ]
[ delivery sem... ]
[ [x] answered ] ◀── card greys out, progress bar ticks up,
[ [View answer] ] and a View answer button appears
At a glance:
| Step | You say / do | What happens |
|---|---|---|
| 1 | prepme — CV: resume.pdf JD: job.txt |
Generates interview-prep.html |
| 2 | Click Copy prompt on a card | Deep-dive prompt copied to clipboard |
| 3 | Paste into your AI agent, work the answer out | You get a strong, follow-up-proof answer |
| 4 | log this answer |
anslog saves the answer page + links it |
| 5 | Reload interview-prep.html |
Card shows answered + View answer; progress advances |
Repeat 2–5 until the progress bar is full. That's the whole game.
Use the skills CLI:
npx skills add pplam/prepmeThis follows the standard skills install flow and lets the CLI guide skill and agent selection interactively.
Or, for a local install straight from a clone:
./install.shCopies the skill into ~/.claude/skills/ (and ~/.codex/skills/ if Codex is set up).
The installable skill lives under skills/prepme/, so the repository README stays at the repo root and is not part of the installed skill payload.
| File | Role |
|---|---|
skills/prepme/SKILL.md |
The brains — how questions are designed and the HTML is assembled. |
skills/prepme/assets/template.html |
The self-contained, light-themed study-sheet template. |
skills/anslog/SKILL.md |
The answer log — saves a worked-out answer back into the study sheet. |
skills/anslog/assets/answer.html |
Template for a saved-answer page (anslog fills it in). |
install.sh |
Local installer — copies both skills into your skills directory. |
Good luck. Go get the offer — then come back and star it out of gratitude.