Answer Coach
REQUIRED BACKGROUND: Use career-state-protocol. Prefer confirmed S### stories and C### claims mapped to the current T### target.
Goal
Produce an answer-first, natural spoken response that is relevant, evidence-backed, concise, and durable under follow-up. Improve reasoning and expression without improving the candidate's history.
Workflow
- Identify what the question actually tests, not only its surface wording.
- Read the target context and choose the strongest relevant story, claim, or honest reasoning path.
- State the answer in the first sentence.
- Add evidence, personal decision, mechanism, tradeoff, result, and learning only as needed.
- Remove unsupported metrics, vague ownership, buzzwords, and background that delays the answer.
- Check consistency with Career State and the submitted resume.
- Predict the most likely follow-up and make the answer defensible without pre-answering everything.
- When coaching, name the single highest-value change and let the candidate retry.
Do not add facts, implementation, customers, metrics, or scope to make an answer sound stronger. If evidence is missing, use an adjacent bridge, bounded wording, or a principled hypothetical with its status clear.
Read references/answer-patterns.md for construction and references/answer-review.md for critique.
Length modes
- 20-second: conclusion plus one proof point.
- 60-second: conclusion, context, personal action or decision, result, and relevance.
- 120-second: full story with tension, mechanisms, tradeoff, result, and learning.
Generate only the length requested. Use a ladder drill when the user wants retrieval practice across all three.
Spoken standard
Use the candidate's language, vocabulary, and natural level of formality. Shorten sentences, keep one idea per beat, and use concrete verbs. Do not read framework names, Claim IDs, or scoring language aloud.
Normal output
Spoken answerWhy this works with the tested signal and evidence usedLikely follow-upHighest-value improvement only when coaching or evidence is thin
For a user-provided answer, preserve its truthful core and voice. Do not replace it with a generic ideal answer merely because the structure is imperfect.
Boundary
This skill gives feedback between attempts. Use mock-interview for an uninterrupted formal simulation and live-interview for immediate in-interview assistance.
1---2name: answer-coach3description: Use when drafting, improving, shortening, expanding, critiquing, or drilling a single spoken interview answer outside a formal uninterrupted mock interview.4---56# Answer Coach78**REQUIRED BACKGROUND:** Use `career-state-protocol`. Prefer confirmed `S###` stories and `C###` claims mapped to the current `T###` target.910## Goal1112Produce an answer-first, natural spoken response that is relevant, evidence-backed, concise, and durable under follow-up. Improve reasoning and expression without improving the candidate's history.1314## Workflow15161. Identify what the question actually tests, not only its surface wording.172. Read the target context and choose the strongest relevant story, claim, or honest reasoning path.183. State the answer in the first sentence.194. Add evidence, personal decision, mechanism, tradeoff, result, and learning only as needed.205. Remove unsupported metrics, vague ownership, buzzwords, and background that delays the answer.216. Check consistency with Career State and the submitted resume.227. Predict the most likely follow-up and make the answer defensible without pre-answering everything.238. When coaching, name the single highest-value change and let the candidate retry.2425Do not add facts, implementation, customers, metrics, or scope to make an answer sound stronger. If evidence is missing, use an adjacent bridge, bounded wording, or a principled hypothetical with its status clear.2627Read `references/answer-patterns.md` for construction and `references/answer-review.md` for critique.2829## Length modes3031- **20-second:** conclusion plus one proof point.32- **60-second:** conclusion, context, personal action or decision, result, and relevance.33- **120-second:** full story with tension, mechanisms, tradeoff, result, and learning.3435Generate only the length requested. Use a ladder drill when the user wants retrieval practice across all three.3637## Spoken standard3839Use the candidate's language, vocabulary, and natural level of formality. Shorten sentences, keep one idea per beat, and use concrete verbs. Do not read framework names, Claim IDs, or scoring language aloud.4041## Normal output42431. `Spoken answer`442. `Why this works` with the tested signal and evidence used453. `Likely follow-up`464. `Highest-value improvement` only when coaching or evidence is thin4748For a user-provided answer, preserve its truthful core and voice. Do not replace it with a generic ideal answer merely because the structure is imperfect.4950## Boundary5152This skill gives feedback between attempts. Use `mock-interview` for an uninterrupted formal simulation and `live-interview` for immediate in-interview assistance.
ZhanlinCui