interview-debrieflisted
Install: claude install-skill vikas9dev/claude-job-hunt
# Interview Debrief
Turn an interview into a record you can learn from. The value is in the **repeats**: one bad
answer is noise; the same mistake in three interviews is a habit to fix.
**Writing style:** short, plain sentences. Technical terms stay exact. Be honest about weak
answers — a kind but vague debrief teaches nothing.
## Step 0 — Load context (every run)
1. Read **`.claude/skills/_shared/PROFILE_AND_FILTERS.md`**. If it does not exist, stop and
tell the user to run `/setup-profile`. The profile gives the user's level (§1, §2) and stack
(§5) — better answers must fit that level.
2. Read **`.claude/skills/_shared/TRACKER_RULES.md`** (job status values).
The store is `data/interviews_data.json`. Use **only** `python3 scripts/interview_track.py`
(`python` on Windows) from the repo root. Never Read or Edit the JSON directly.
## Privacy rules (always)
- **No interviewer names.** Record the role only: "Senior engineer", "Engineering manager",
"HR". If the user gives a name, drop it.
- **Paraphrase questions.** Keep enough to practise from ("design a URL shortener"). If the
user is under an NDA or the company asked not to share questions, write only the topic and
a one-line paraphrase.
- No salary figures from the interview here — those belong in the job tracker or `/offer-compare`.
---
## Mode A — Debrief (default)
### A1. Collect
Ask once, in one message, and wait:
```
Tell me everything you remember, in any order:
- company, role, and which round(