Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Build a structured, role-specific interview question bank with what good answers look like. Use when asked to create interview questions, an interview guide, a structured interview kit, or competency-based questions for a role. Produces questions mapped to the competencies that matter — behavioral (STAR), role/technical, and values — each with what a strong vs. weak answer shows and follow-up probes, for fair, consistent interviews.
.claude/skills/mohitagw15856-interview-question-bank/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 58% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 80% | 0% |
Unstructured interviews mostly measure who's charming. Structured, competency-based interviews predict performance — the same questions, mapped to what the role needs, scored against what a good answer looks like. This skill builds that question bank so every interviewer assesses the same things, fairly and consistently.
Given "interview questions for a senior PM", build the bank anyway — infer the core competencies for the role and write questions for each, labelling assumptions. Provide "what good looks like" for every question. Never hand back a flat list of questions with no evaluation guidance.
Ask for these only if they aren't already provided (else infer and label):
1. Competency map — the 4–6 competencies to assess and which round/interviewer owns each (avoid everyone asking the same thing).
2. Questions by competency — for each competency, 2–4 questions:
| Question | Type | What a strong answer shows | Red flags | Follow-up probes | |---|---|---|---|---| | "Tell me about a time you…" | Behavioral (STAR) | specifics, their role, the outcome, learning | vague, all "we", no result | "What would you do differently?" |
Include behavioral (past behaviour, STAR-friendly), role/technical (a realistic problem or scenario), and values questions.
3. Scoring — a simple rubric (e.g. 1–4 per competency) and the bar to advance, so scores are comparable across interviewers.
4. Fairness notes — ask every candidate the same core questions; keep questions job-related; avoid questions about protected characteristics (age, family, health, religion, etc.); focus on evidence, not "fit feeling".
Structured-interview practice — competency-based, behaviorally-anchored questions with scoring rubrics and fairness/consistency safeguards.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 24,007 | 28,666 | +19% | 1 | 1 | 0% | 4,687 | 6,229 | +33% | 0 | 0 | — |
case-02 | fail→fail | 20,925 | 17,882 | -15% | 1 | 1 | 0% | 3,799 | 3,921 | +3% | 0 | 0 | — |
case-03 | fail→pass | 23,968 | 23,353 | -3% | 1 | 1 | 0% | 3,890 | 4,655 | +20% | 0 | 0 | — |
case-04 | pass→fail | 12,151 | 26,262 | +116% | 1 | 1 | 0% | 2,125 | 5,266 | +148% | 0 | 0 | — |
case-05 | pass→pass | 7,205 | 8,649 | +20% | 1 | 1 | 0% | 1,254 | 2,193 | +75% | 0 | 0 | — |
case-06 | pass→fail | 20,451 | 25,644 | +25% | 1 | 1 | 0% | 3,153 | 4,583 | +45% | 0 | 0 | — |
case-07 | pass→pass | 18,023 | 20,012 | +11% | 1 | 1 | 0% | 3,138 | 4,183 | +33% | 0 | 0 | — |
case-08 | pass→pass | 20,254 | 22,781 | +12% | 1 | 1 | 0% | 3,601 | 5,009 | +39% | 0 | 0 | — |
case-09 | fail→pass | 16,392 | 21,807 | +33% | 1 | 1 | 0% | 2,857 | 4,519 | +58% | 0 | 0 | — |
case-10 | pass→pass | 16,782 | 20,290 | +21% | 1 | 1 | 0% | 2,979 | 4,017 | +35% | 0 | 0 | — |
case-11 | pass→pass | 15,115 | 18,238 | +21% | 1 | 1 | 0% | 2,621 | 3,948 | +51% | 0 | 0 | — |
case-12 | pass→pass | 20,274 | 19,800 | -2% | 1 | 1 | 0% | 3,474 | 4,255 | +22% | 0 | 0 | — |
case-13 | fail→pass | 15,946 | 18,224 | +14% | 1 | 1 | 0% | 2,862 | 4,080 | +43% | 0 | 0 | — |
case-14 | fail→pass | 15,393 | 26,140 | +70% | 1 | 1 | 0% | 2,784 | 5,014 | +80% | 0 | 0 | — |
case-15 | pass→pass | 15,223 | 20,557 | +35% | 1 | 1 | 0% | 2,782 | 4,261 | +53% | 0 | 0 | — |
case-16 | pass→pass | 19,249 | 23,580 | +22% | 1 | 1 | 0% | 3,589 | 4,834 | +35% | 0 | 0 | — |
case-17 | pass→pass | 19,796 | 23,040 | +16% | 1 | 1 | 0% | 3,429 | 4,612 | +34% | 0 | 0 | — |
case-18 | fail→pass | 16,754 | 19,416 | +16% | 1 | 1 | 0% | 2,925 | 4,148 | +42% | 0 | 0 | — |
case-19 | fail→pass | 18,993 | 23,732 | +25% | 1 | 1 | 0% | 3,272 | 4,746 | +45% | 0 | 0 | — |
case-20 | fail→pass | 26,830 | 19,836 | -26% | 1 | 1 | 0% | 4,933 | 4,154 | -16% | 0 | 0 | — |
case-21 | pass→pass | 16,094 | 16,990 | +6% | 1 | 1 | 0% | 2,771 | 3,526 | +27% | 0 | 0 | — |
case-22 | pass→pass | 19,433 | 20,401 | +5% | 1 | 1 | 0% | 3,665 | 4,120 | +12% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +27 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.