cover-letter

name	cover-letter
description	Korean cover-letter writing pipeline. Drafts an applicant cover letter (자기소개서) in STAR (Situation/Task/Action/Result) form, then runs deterministic AI-tell scrubbing. Use when the user asks for a cover letter, 자기소개서, or career transition statement. Argument "review" runs only the AI-tell scrubber on an existing file.

Cover-letter Pipeline (Orchestrator)

Pipeline overview

Phase 0  Inputs collection            [main context, interactive]
Phase 1  Skeleton framing             [main context, interactive]
Phase 2  aiwriting-writer subagent    [fresh context, drafts to file]
Phase 3  aiwriting-scrubber subagent  [fresh context, R1-R7 deterministic]
Phase 4b aiwriting-copy-killer        [LLM-free score, always]
Phase 4c aiwriting-fact-checker       [LLM-free pattern, always]

The structure-critic (Phase 4) is not routed for cover-letter by default. The argumentation rigor demanded for blog/paper exceeds what hiring readers expect; STAR clarity matters more.

Knowledge files (loaded by subagents, not orchestrator), under ${CLAUDE_PLUGIN_ROOT}/skills/cover-letter/:

• philosophy.md - shared with blog/paper (Orwell/Zinsser/Graham/Popper)
• cover-letter-templates.md - STAR template + opening hook patterns + closing patterns

The orchestrator does NOT load these.

---

Phase 0: Inputs (Interactive)

If $ARGUMENTS is review [path], jump to Phase 3 (scrub-only) on the given file.

Otherwise, collect:

• 지원 회사 / 직무 / 연차
• 한 줄로 본인의 강점 (왜 합격해야 하는가)
• 가장 단단한 사례 1-2개 (수치 가능하면 수치)
• 톤: ~습니다 (default for cover-letter) or ~다

If the user pasted a job description, skim and ask:

지원 동기를 회사 측 어떤 한 줄(JD 발췌)에 묶을까요?

Phase 1: Skeleton Framing (Interactive)

1. Generate a 5-section skeleton: intro / experience1 / experience2 / fit / closing.
2. Each line is the topic sentence the section will defend (STAR shape). Show to user, iterate until confirmed.
3. Confirm tone. Default ~습니다.

Phase 2: Spawn aiwriting-writer

Agent({
  subagent_type: "aiwriting-writer",
  description: "Draft cover-letter for {role} @ {company}",
  prompt: "
    Format: cover-letter
    Topic: {role} @ {company}
    Skeleton:
      intro: {one-liner}
      experience1: {one-liner}
      experience2: {one-liner}
      fit: {one-liner}
      closing: {one-liner}
    Tone: {~습니다 / ~다}
    Output target: cover-letter-drafts/{kebab-case-slug}.md (current working directory)

    Load knowledge from skills/cover-letter/{philosophy,cover-letter-templates}.md
    (plugin-relative). Return the absolute path of the saved file plus a short summary.
  "
})

Capture the returned absolute path.

Phase 3: Spawn aiwriting-scrubber

Agent({
  subagent_type: "aiwriting-scrubber",
  description: "Scrub AI tells from {filename}",
  prompt: "
    Draft file: {absolute_path_from_phase_2}
    Format: cover-letter

    Apply R1-R7 from skills/blog/ai-tell-rules.md (plugin-relative).
    Cover-letter does NOT require a `## 요약` section - skip the R3 summary insertion.
    Return the scrub report.
  "
})

If $ARGUMENTS was review [path], stop here and surface the report.

Phase 4b: aiwriting-copy-killer (LLM-free)

Agent({
  subagent_type: "aiwriting-copy-killer",
  description: "Score AI-likeness of {filename}",
  prompt: "Draft file: {absolute_path}. Threshold: 0.35."
})

Phase 4c: aiwriting-fact-checker (LLM-free)

Agent({
  subagent_type: "aiwriting-fact-checker",
  description: "Fact-check {filename}",
  prompt: "Draft file: {absolute_path}. Known facts: known_facts.yml (cwd)."
})

Hard-evidence categories (numbers, dates, semver, quotes, proper nouns) are flagged when not in known_facts.yml. For cover-letter, this catches inflated experience claims and unverified company facts.

Decision matrix

Scenario	writer	scrubber	critic	copy-killer	fact-checker
Standard cover-letter	yes	yes	skip	yes	yes
/aiwriting:cover-letter review {path}	skip	yes	skip	yes	yes
User insists on argumentation review	yes	yes	optional	yes	yes

Important notes

• 자기소개서 첫 문장은 가장 단단한 단언으로. "안녕하세요" 류 인사 금지.
• "최선을 다하겠습니다" 류 추상 수식어 0회 (R2 + ai-tell-rules.md 카탈로그).
• 회사명·직무명은 본문 어딘가 1회는 명시.
• 수치는 known_facts.yml 의 항목으로 yaml 등록. fact-checker 가 통과시킨다.