Structured interview loop builder with Claude

A Claude Skill that takes a job description, the company’s structured interviewing standards, and produces a complete interview loop design — interview stages, per-stage rubric dimensions, behavioral questions per dimension, scorecard templates, and recommended interviewer assignments. Replaces “we’ll figure out the interview process when we have a candidate” with a 30-minute setup that produces operational discipline.

What you’ll need

• Claude Code or Claude.ai with custom Skills enabled
• The role’s job description (output from the JD writer Skill or equivalent)
• The company’s interview standards — typical loop length, required interview types, interviewer-pool constraints
• The team’s structured interview rubric library

Setup

1. Drop the Skill. Place interview-loop-builder.skill into your Claude Code skills directory. The Skill exposes one callable function: design_loop.
2. Configure the standards. Edit interview_standards.yaml with: maximum loop length (e.g., 4 stages), required interview types (recruiter screen, HM screen, on-site loop), interviewer-pool eligibility per role family.
3. Configure the rubric library. Place existing role-family rubrics in rubrics/ for the Skill to pattern-match against.
4. Test on a known role. Run on a role where the loop design already exists. Compare the Skill’s output to the existing loop; tune the config.

How it works

The Skill takes the JD and:

1. Identifies the rubric dimensions. From the JD’s must-have-skills section, maps to the team’s rubric library — what’s being evaluated, at what depth.
2. Designs the stage progression. Recruiter screen (fit + interest), HM screen (depth on top dimensions), on-site loop (full rubric coverage with one dimension per interviewer where possible).
3. Generates per-stage interview content. Behavioral questions per rubric dimension, suggested probing follow-ups, scorecard with anchor descriptions per score level.
4. Suggests interviewer assignments. Based on rubric-dimension-to-interviewer-skill matching from the eligible interviewer pool.

Output

A complete interview loop document with:

# Interview Loop: [Role] @ [Team]

## Stage 1: Recruiter Screen (30 min)
- Goal: confirm fit, interest, basic qualifications, comp alignment
- Key questions: [3-5 questions]
- Disqualifying signals: [list]

## Stage 2: Hiring Manager Screen (45 min)
- Goal: depth on top 2 rubric dimensions
- Rubric dimensions covered: [list]
- Behavioral questions: [4-6 questions with rationale]
- Scorecard template: [linked]

## Stage 3: On-Site Loop (4 interviews, 60 min each)
- Interview A — [Dimension 1, e.g. Technical Depth]
  - Questions: [3-4 behavioral + 1 technical exercise]
  - Suggested interviewer: [pool match]
- Interview B — [Dimension 2, e.g. Systems Design]
  - ...
- Interview C — [Dimension 3, e.g. Collaboration]
  - ...
- Interview D — [Dimension 4, e.g. Leadership / Influence]
  - ...

## Stage 4: Debrief
- Format: independent scoring before group discussion
- Rubric: [linked]
- Decision criteria: [explicit thresholds]

Where it fits

Use this Skill at the start of every new role — alongside the JD writer. The output gets configured in Ashby or Greenhouse as the role’s interview structure; recruiters and hiring managers run candidates through the resulting structured loop.

The compounding benefit: structured loops produce better quality of hire than ad-hoc loops, with materially less per-role design overhead when AI handles the design.

Watch-outs

• Rubric quality determines loop quality. A vague rubric library produces vague interview questions. Invest in the rubric design before deploying the Skill at scale.
• Hiring manager review is mandatory. AI-designed loops are the starting point; the hiring manager validates the dimensions reflect the actual role priorities.
• Calibrate against your culture. Loop length, interview style, behavioral-vs-technical balance vary by company. Configure the standards file to match.
• Don’t skip interviewer training. The loop is operational, not magical; interviewers still need training on behavioral interviewing discipline and rubric application.
• Update with feedback. When interview intelligence reveals questions that aren’t producing useful signal, update the Skill’s config.