| name | interview-storytelling |
| description | Analyze interview transcripts to evaluate which stories were chosen, which should have been chosen, and how to reframe answers for next time. Use after a real interview to turn the experience into structured feedback and new STARR achievements. |
| version | 0.1.0-stub |
| status | placeholder — to be filled in a dedicated session |
| author | Job Hunt System |
| tags | ["interview","storytelling","retrospective","behavioral","STARR"] |
Interview Storytelling Skill (Placeholder)
⚠️ Status: STUB. This skill is being designed in a dedicated session. The structure below is a placeholder that captures intended scope. Do not rely on the details — they will be revised.
🎯 Purpose
Help the user turn an interview (especially behavioral / experience-based ones) into a structured retrospective:
- Which stories did I pick? Were they the strongest available for each question?
- Which stories should I have picked? What did my story library have that I didn't use?
- What didn't land? Where did the interviewer push back, lose interest, or change topic?
- How do I retell the weak ones? Reframe, restructure, add metric, change opening.
- What gaps does this expose? Questions I had no story for — feed back into
/add-achievement.
The goal is twofold:
- Get better at the next interview.
- Surface forgotten achievements that should have been in the story library all along.
🧩 Inputs (intended)
- Interview transcript (pasted text, or file in
interviews/)
- Target role context (
target_roles/[role]/role_profile.md)
- Available story library (
achievements/my_data/)
- Optional: JD for the company being interviewed at
🔄 Intended Workflow (rough sketch)
- Parse the transcript. Identify each question + the story the user used to answer.
- Match stories to library. For each used story, find the closest match in
achievements/my_data/.
- Score the choice. Was the chosen story the strongest available for that question, given role priorities?
- Suggest alternatives. For weak choices, recommend swaps from the library.
- Surface gaps. Questions where the user had no good story — flag for
/add-achievement.
- Reframe weak answers. For stories that didn't land, draft alternative openings, structure, or framing.
- Save retrospective. Output to
interviews/interview_[company]_[date].md.
📐 Intended Output (placeholder format)
# Interview Retrospective — [Company] — [Date]
## Questions & Story Choices
### Q1: "Tell me about a time you led a difficult project."
- **Story used:** story_data_framework
- **Match score:** ✅ Strong — Tier 1 for this role
- **What landed:** clear metric, owned the action
- **What to improve:** opening was too long, get to the result faster
### Q2: "Walk me through a conflict with a stakeholder."
- **Story used:** story_pricing_dispute
- **Match score:** ⚠️ Weak — story exists in library but not the strongest available
- **Better alternative:** story_legal_negotiation (Tier 1 for conflict + influence)
- **Why better:** higher stakes, clearer outcome, role-aligned
## Gaps Surfaced
- No story for: "tell me about a failure"
- Suggested: extract from your time at [Company X] — you mentioned the [Y] rollback
- Run `/add-achievement` to capture it.
## Reframing Notes
- ...
## Next Steps
- [ ] Add 1–2 missing stories via `/add-achievement`
- [ ] Practice reframed opening for story_pricing_dispute
- [ ] Re-run `/map-skills` if new stories are added
🔗 Integration with Other Skills
starr-achievements — gaps surfaced here feed /add-achievement
job-analysis — role priorities determine which stories were "strongest available"
applications — interview retrospectives link back to the application that produced them
fix-errors — if the retrospective reveals a factual error in a story, propagate via /fix
📁 Intended File Locations
- Skill:
.claude/skills/interview-storytelling/SKILL.md (this file)
- Command:
.claude/commands/analyze_interview.md (stub)
- Output folder:
interviews/my_data/interview_[company]_[date].md (folder TBD)
- Reference templates:
.claude/skills/interview-storytelling/references/ (TBD)
📝 Open Questions (to resolve in the dedicated session)
- Format of transcript input — paste, file, audio + transcription?
- How to score "strongest available story" — heuristic, LLM judgement, or user-confirmed?
- How to handle interviews that aren't fully STARR-style (e.g. case studies, system design)?
- Do retrospectives live under
interviews/ (new folder) or inside applications/app_*/?
- How aggressive should reframing suggestions be vs honest acknowledgement of weak fit?
TODO before promoting from stub to v1.0: