last update
Apr 17, 2026
license
MITv1.0.0
format
SKILL.md
- Install
// Downloads SKILL.md into ~/.claude/skills/grill-me-interview-prep/
$ curl -fsSL --create-dirs -o ~/.claude/skills/grill-me-interview-prep/SKILL.md https://claudeskil.com/skills/grill-me-interview-prep/SKILL.md - Invoke
// Inside Claude Code, type the skill name as a slash command
> /grill-me-interview-prep help me ship this - Or copy this from the README
// extracted from a Usage / Example section
$ curl -fsSL --create-dirs -o ~/.claude/skills/grill-me-interview-prep/SKILL.md https://claudeskil.com/skills/grill-me-interview-prep/SKILL.md > Grill me. System design, staff level. Topic: design a rate limiter. [Interviewer] You have one prompt. Walk me through the API first, then storage. I'll interrupt.
- name:
- Grill Me Interview Prep
- slug:
- grill-me-interview-prep
- version:
- v1.0.0
- license:
- MIT
- author:
- @chiragthummar16
- categories:
- tags:
- #interview#prep#system-design#behavioral#coaching
- description:
Mock interviews on demand. System design, behavioral, coding — with brutal-honest feedback.
README.md
Grill Me Interview Prep
A mock interviewer that does not go easy on you. Grill Me Interview Prep runs realistic interview rounds — system design, behavioral, and coding — playing a skeptical senior interviewer who probes weak answers instead of nodding along. Built for engineers targeting senior and staff roles who want pressure-tested practice before the real loop, not a checklist of canned questions.
When to use it
- You have an onsite or virtual loop in the next week and want reps against a tough interviewer.
- You keep freezing on system design and need someone to ask "why not just use a queue here?" until you can defend a choice.
- Your behavioral stories ramble; you want STAR-shaped feedback on a specific story.
- You bombed a recent round and want a debrief on what actually went wrong.
How it works
The skill runs a stateful interview session inside the conversation. You pick a track (system-design, behavioral, or coding) and a difficulty target (senior, staff). It then:
- Opens with one prompt, not a wall of questions.
- Listens to your answer and fires follow-ups on the weakest point — vague tradeoffs, hand-waved scaling numbers, missing failure modes.
- Holds the interviewer persona: no hints unless you explicitly ask, no premature praise.
- Ends the round with a written debrief: a score per dimension (clarity, depth, tradeoff reasoning, communication), the two or three things to fix first, and a model answer for comparison.
You drive pacing in plain language — "move on", "give me a hint", "end the round and debrief me", or "harder, this was too easy".
Example
curl -fsSL --create-dirs -o ~/.claude/skills/grill-me-interview-prep/SKILL.md https://claudeskil.com/skills/grill-me-interview-prep/SKILL.md
> Grill me. System design, staff level. Topic: design a rate limiter.
[Interviewer] You have one prompt. Walk me through the API first,
then storage. I'll interrupt.
> ...sliding window with Redis...
[Interviewer] Redis goes down mid-window. What does a client see,
and how do you keep that from cascading? Be specific about timeouts.
> end round, debrief me
Tips & gotchas
- Name the role and level up front. "Staff at a payments company" produces sharper, domain-specific grilling than a bare topic.
- It will not volunteer hints. That is intentional. Say "hint" when you're truly stuck so the debrief still reflects real difficulty.
- Paste the real job description if you have one — it tailors follow-ups toward what that team actually screens for.
- Ask for the model answer at the end; the debrief withholds it by default so you attempt the question cold.
FAQ
Can it do live coding with a specific language? Yes. Specify the language and it grades on correctness, edge cases, and how you communicate while coding — not just whether it compiles.
Will it stay in character if I argue? Yes — push back on a tradeoff and it pushes back harder, the way a real interviewer would. Say "break character" to step out and discuss.
Can I run several short rounds in one session? Yes. Each round keeps its own debrief, and you can ask for a combined summary of recurring weak spots across all of them.
$ cat reviews/