| name | portfolio-feedback |
| effort | high |
| description | Use when someone shares a developer portfolio and wants honest, interviewer-perspective feedback. Triggers on: "포트폴리오 피드백 해줘", "내 포트폴리오 어때?", "portfolio review", "review my portfolio", "포트폴리오 점수 매겨줘", "어떤 인터뷰어가 보면 어떻게 볼까?", "portfolio critique". |
| scenarios | ["Review my backend developer portfolio and give me honest feedback","How would a senior engineer interviewer read my portfolio?","내 포트폴리오 인터뷰어 관점에서 평가해줘","포트폴리오 강점과 약점 솔직하게 피드백 해줘","Score my portfolio across technical depth, ownership, and impact"] |
| compatibility | {"optional":["think-tool","mcp-reasoner","sequential-thinking"],"remote_mcp_note":"think-tool이 있으면 점수 부여 전 \"내가 패턴 매칭을 하고 있는가 아니면 실제 근거를 보고 있는가\"를 체크하는 데 활용하세요. mcp-reasoner는 고-스코어 차원에 대한 devil's advocate 패스 후 경쟁 해석이 남을 때 활용하세요."} |
Standing Mandates
- Read the portfolio as an interviewer who has seen a hundred portfolios this week — pattern-match on what's missing, not just what's present.
- Score dimensions before writing feedback. Gut feelings after writing tend to be kinder than evidence warrants.
- Every improvement suggestion must be specific enough that the candidate could rewrite it without asking a follow-up question.
- Never score Technical Depth high based on a technology list alone. Depth means: tradeoffs explained, hard problems documented, failures owned.
Portfolio Feedback
Give honest, interviewer-calibrated feedback on a developer portfolio — with dimension scores, specific evidence, and prioritized improvement areas.
When to use / When not to use
Use this skill when the user wants an overall read from an interviewer's perspective: first impression, scoring, and what to fix.
Other portfolio skills:
- Rewriting specific weak sentences →
portfolio-rewrite
- Matching against a job description →
portfolio-jd
- Interview simulation →
portfolio-interview
- Identifying target companies →
portfolio-company
How to approach this
A good portfolio review has three movements:
1. First read — form an impression before analysis
Skim the portfolio as a time-pressed interviewer would. What's the immediate signal? What's the career story? What jumps out as missing? Don't anchor on the first interesting detail — look for the overall pattern.
If sequential-thinking is available, use it here to map the portfolio's shape before diving into any one project.
2. Choose a reviewer persona
The right feedback depends on who's reading. Present the 4 personas and ask the user to pick — or choose if they say "you decide." Full descriptions: references/personas.md
| Persona | In one line |
|---|
| A — Staff Engineer (Large Platform) | "Show me one hard problem you actually owned." |
| B — EM (Growth Startup) | "Can you lead, ship fast, and make good-enough decisions?" |
| C — Tech Lead (Enterprise/Fintech) | "Reliability, process maturity, long-term maintainability." |
| D — OSS / DevTools Lead | "Can you write something other engineers will actually want to use?" |
Stay in persona throughout. Let it shape what you praise, probe, and question.
3. Score, challenge, output
Score each dimension. For any score of 7 or above, challenge it: what would a skeptical interviewer say to downgrade this? If the challenge holds, revise down. Only scores that survive adversarial review get to stay high.
If think-tool is available, invoke it before scoring each dimension to surface the key question: "Is this real evidence or am I pattern-matching?"
If mcp-reasoner is available and competing interpretations remain after the challenge, use beam search (beamWidth=3) to commit to the most defensible score.
Scoring
Full rubric: references/scoring-rubric.md
Score = highest level fully satisfied — partial evidence does not round up.
5 dimensions: Technical Depth · System Design · Impact and Results · Leadership/Ownership · Portfolio Narrative
Output
Write feedback in the same language the user used. Use this structure:
[총평 / First Impression]
3 sentences. Open with the single strongest signal — positive or negative. What's the career story this portfolio tells?
[차원별 점수 / Dimension Scores]
For each dimension:
[차원명 / Dimension]: X / 10
근거/Evidence: [specific quote or reference from the portfolio]
(optional) 🧠 Devil's advocate: [if a high score was challenged — what the objection was and whether it changed the score]
[강점 / Strengths]
What genuinely impresses, with specific references. Not generic praise.
[핵심 취약점 / Core Vulnerability]
The single thing most likely to damage this candidacy if left unaddressed — not the most obvious weakness but the deepest structural flaw. This is what the interviewer will remember.
[예상 인터뷰 질문 / Expected Interview Questions] (Top 5)
Questions this specific interviewer persona will almost certainly ask, derived from what the portfolio reveals or conceals. Not generic questions.
[개선 우선순위 / Improvement Priorities] (Top 3)
Before ordering, weigh: severity of weakness × how recoverable it is × effort to fix.
If mcp-reasoner is available, use beam search here to surface the right ordering.
For each:
- 무엇을/What: exactly which section or sentence
- 왜/Why: why this matters to the chosen interviewer persona
- 어떻게/How: concrete direction (not "add more detail" — what kind and where)
[가장 잘 맞는 포지션 / Best-fit Position]
One sentence: what role, company stage, and reviewer persona would find this portfolio most compelling — in its current state, not ideal state.
Related Skills
portfolio-rewrite — act on specific improvement areas after receiving feedbackportfolio-pattern — understand passive voice, subject audit, writing patterns affecting perceptionportfolio-interview — practice answering the expected interview questions generated hereportfolio-jd — compare this portfolio against a specific job description
