Behavioral Interview Prepper

Generate a tailored behavioral interview answer bank from resume-builder output. Output goes to the same hojicha/<company>-<role>-resume/ directory as behavioral-prep.md.

Critical Rules

1. NEVER fabricate experiences. Only use content from the resume and candidate-context.md. Rephrase and reframe — never invent.
2. Chain from resume-builder output. Read existing notes.md and resume.tex from the output directory. Do not re-parse the JD from scratch.
3. Honest gap handling. When the candidate lacks an experience for a question and has confirmed this after probing (see Rule 5 and Step 4 inline probing), provide a deflection strategy — not a made-up story.
4. Story reuse limit. No single experience may be used for more than 5 questions. Uses 4-5 require a distinct reframing angle (different trait cluster emphasis). See references/story-mapping.md.
5. Never generate generic or hypothetical answers — ask instead. If you would produce a vague behavioral answer, stock STAR response, or hypothetical framework where a real experience should be, STOP and ask the candidate a probing question. See references/candidate-discovery.md for anti-generic detection and probing techniques. Every answer must be grounded in real experience. Hypothetical frameworks are only acceptable when the candidate explicitly confirms they have no related experience.

Workflow

Step 1: Parse Inputs

Read from the existing resume-builder output directory: