# Story Bank Builder

> Use when mining, building, auditing, prioritizing, or updating a candidate's reusable interview stories from confirmed career experiences and project dossiers.

- Skill: `zhanlincui/story-bank-builder` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add zhanlincui/story-bank-builder`
- Raw SKILL.md: https://api.skillmd.com/api/skills/zhanlincui/story-bank-builder/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: ZhanlinCui (https://skillmd.com/u/zhanlincui)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/zhanlincui/story-bank-builder

---


# Story Bank Builder

**REQUIRED BACKGROUND:** Use `career-state-protocol`. Use `project-deep-dive` when a source project is too thin to support a story.

## Goal

Build a compact portfolio of distinctive, evidence-backed stories that covers the candidate's real interview needs. Optimize for coverage and defensibility, not a fixed story count.

## Inputs

Read the Career Profile, `P###` dossiers, confirmed and bounded `C###` claims, existing `S###` stories, target models when available, and prior interview use or feedback. Do not invent missing project detail in the story layer.

## Workflow

1. Build a coverage matrix from the supplied `T###` model. Without a target, use broad career signals such as ownership, ambiguity, product or technical judgment, evaluation, collaboration, conflict, failure, customer or user insight, measurable impact, and learning.
2. Inventory candidate events that contain a real tension, decision, action, result, or changed operating method.
3. Score each event for relevance, evidence, ownership, distinctiveness, follow-up safety, and overlap with other stories.
4. Select the smallest portfolio that provides strong coverage and useful backups.
5. For thin but promising events, create a `G###` item and route to `project-deep-dive` before drafting.
6. Draft each story from linked claims, then test skeptical follow-ups.
7. Ask the user to confirm any new framing that changes emphasis, causality, or personal attribution.
8. Update story strength, follow-up risks, use count, and interview feedback after reuse.

Read `references/coverage-and-selection.md` before selecting stories and `references/story-contract.md` before writing them.

## Story construction

Prefer a decision narrative over mechanical STAR labels in spoken output:

- context and stakes;
- exact responsibility;
- hard tension or decision;
- specific actions and mechanisms;
- evidence-supported result and attribution;
- learning that changed later behavior.

Prepare a 30-second version for retrieval, a 90-second version for normal answers, and deep-dive bullets for follow-up. All versions must use the same claims and boundaries.

## Portfolio rules

- One strong story may answer several questions, but do not pretend reuse creates broader evidence.
- Track follow-up risks and use count to prevent repetitive interviews.
- Preserve honest adjacent bridges where a perfect story does not exist.
- A failure story needs a real miss and changed method, not a disguised success.
- A conflict story needs a real disagreement and resolution mechanism, not "we communicated."
- Target-specific mappings belong under `T###`; the base story remains target-neutral.