agentleFS
Hrutin-Suriya-KotniHrutin-Suriya-Kotni/until-hired/skills/interview-coach/SKILL.md

interview-coach

Skill2 starsChanged 3 days agoMITView on GitHub

Interview preparation, mock interviews and same-day debriefs. Builds a 15-minute prep sheet for one round from the candidate's real work, the job, company research and their known gaps; runs realistic mock interviews with honest scoring; turns every debrief into tracked gaps so the same weakness never costs two interviews. Use when an interview or assessment is scheduled, when the candidate says 'prep me for <company>', 'mock interview', 'grill me on <topic>', or 'debrief'.

17 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 4 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 4

    Distinctive

    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:

    Write <folder>/prep_<round>.md (the job's folder from tracker.py folder --id N), readable in 15 minutes:

    line 14
  2. Distinctive:

    Write talking points as short spoken phrases, not paragraphs.

    line 31
  3. Distinctive:

    Ask one question at a time, like a real interviewer.

    line 35
  4. Distinctive:

    After each answer: a score out of 5 for content, structure and evidence.

    line 36
  5. Distinctive:

    Ask for three lines: what went well, what they probed, what didn't go well.

    line 40
  6. Very specific:

    Every weak spot becomes a gap: {"topic": "<topic>", "source": "interview <Company> <date>", "note": "<what was asked>", "skill": "<skill, …

    line 42

“Common” is measured: kept by at least 6 in 10 of the 11 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

Interview coach

A. Prep sheet (one per round)

Inputs: the tracker row and interview (tracker.py show --id N), the JD, the invite email (round type, duration, interviewers), FACTS.md, STORIES.md, DO_NOT_CLAIM.md, knowledge/GAPS.md, and past debriefs for this company. Research (a researcher subagent if available): what the company sells and to whom, recent news, tech or product details from their own pages, the interviewers' public roles. Cite every fact with a URL. Reuse research under 30 days old.

Write <folder>/prep_<round>.md (the job's folder from tracker.py folder --id N), readable in 15 minutes:

# <Role> at <Company>: <round>, <date and time, candidate's timezone>
## The role in 3 lines
## Start here: your known gaps that this round may hit (from GAPS.md and past debriefs)
## Likely questions, most likely first
   Behavioural (5-7)  → which story from STORIES.md, and the one-line point
   Technical (5-7)    → talking points with the real numbers from FACTS.md
   Design / case (2-3)→ an approach outline and which real project shows it
   Company (2-3)      → what to know; how their business works
## Honest answers for weak spots (closest real experience + how you'd close the gap)
## 3-5 points to work in anywhere
## Questions to ask them (specific to this team)
## Pay: your stance if it comes up (from PREFERENCES.md; never state the private floor)
## Logistics: link, people, what to have ready
## Sources

Write talking points as short spoken phrases, not paragraphs. Notify the candidate (high) when the sheet is ready, and add it to the tracker note.

B. Mock interview ("mock interview", "grill me on <topic>")

  • Agree the format: round type, length, interviewer style (friendly, neutral, tough).
  • Ask one question at a time, like a real interviewer. Follow up on vague answers ("What exactly did you do?", "Why that design?", "What would break at 10x?").
  • After each answer: a score out of 5 for content, structure and evidence; one thing that worked; one concrete fix; a better version in their own words (only real facts).
  • End with the top 3 fixes and update STORIES.md if a better story emerged.

C. Debrief (same day, after every round)

  1. Ask for three lines: what went well, what they probed, what didn't go well. Plus a self-rating out of 10 and the stated next step.
  2. Record: write {"went_well": "…", "probed": "…", "struggled": "…", "rating": "7/10", "outcome": "…"} to a fresh input file (agent-setup references/safe-commands.md), then tracker.py debrief --id <interview id> --json <path>.
  3. Every weak spot becomes a gap: {"topic": "<topic>", "source": "interview <Company> <date>", "note": "<what was asked>", "skill": "<skill, optional>"} → knowledge.py gap --json <path>.
  4. Feedback that something was strong → note it in PROFILE.md (strengths with proof).
  5. Draft a short thank-you note for the candidate to send if they want one (2–4 sentences, one specific thing from the conversation).
  6. Set the next step in the tracker (--next … --next-on …) so silence gets noticed.

Rules

  • Real experience only: every answer points at something in FACTS.md. Never coach the candidate to claim what they haven't done.
  • Gaps are said honestly: the closest real thing, and how they'd learn the rest. No "I'm a quick learner".
  • Most likely questions first in every group.
  • Assessments and take-home tasks: help the candidate prepare and review their own work; don't do graded work for them.

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.