Files

T

Antigravity Agent c24165b8bc refactor(topics): 멀티 에이전트용 지식 재편 — _Common(공통 기본기) + Domain_* 구조

에이전트 8종(대화형/프로그래머 C·S/디자이너/설계자/기획자/QA/PD/PM)에게
[공통 기본 능력 + 롤별 Specialty] 2층으로 지식을 주입하기 위한 재분류.
문서 내용·포맷은 무수정, 폴더 이동만 (6,372개 문서 수 보존 확인).

- Topic_Programming → Domain_Programming (내부 구조 보존)
- Topic_Graphic → Domain_Design
- Topic_Business → Domain_Product
- Topic_General → Domain_General
- _Common 신설: Math(구 Topic_Math_Specialty), Reasoning(구 General/From_Thinking & Reasoning),
  Reasoning_Creativity(구 General/From_창의성), Communication(Poetic_Blog_Writing + From_writing)
- 타 도메인의 From_* 폴더는 유지 (출처 표기일 뿐, 이미 도메인에 맞게 분류된 문서)
- 빈 폴더 정리 (memory/procedures)
- 에이전트→폴더 매핑은 workspace의 .astra/agent-knowledge-map.json (9개 에이전트)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

2026-07-11 11:05:56 +09:00

4.9 KiB

Raw Blame History

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack

title

Articulateness

📌 한 줄 통찰

"매 thought 의 high-resolution output". 매 head 의 fuzzy idea 의 매 lossless 의 transmit. 매 AI 시대 의 가장 큰 leverage — 매 prompt 의 articulate = 매 output 의 quality. 매 vocabulary 의 X, 매 clarity 의 win.

📖 핵심

매 components

Vocabulary precision: 매 정확 단어 선택. 매 vague vs specific.
Structural clarity: 매 conclusion-first (BLUF, Bottom Line Up Front).
Nuance: 매 audience 의 calibrate.
Cohesion: 매 logical flow 의 transition.
Concision: 매 superfluous word 의 cut.

매 BLUF (Bottom Line Up Front)

❌ "Background, then build-up, then conclusion."
✅ "Conclusion. Then evidence."

→ 매 reader 의 시간 존중.

Plain Language (매 modern standard)

매 short sentence (15-20 word).
매 active voice ("X did Y" > "Y was done by X").
매 concrete > abstract.
매 jargon 의 explain.
매 multinational 의 translation 의 friendly.

→ US Plain Writing Act (2010), GOV.UK style guide.

매 prompt engineering 의 articulateness

Specific: 매 "write a poem" → "매 4-line haiku 의 autumn".
Constraint: 매 length, 매 style, 매 audience.
Example: 매 few-shot.
Role: 매 "act as senior backend engineer".
Output format: 매 JSON, 매 bullet, 매 markdown.

매 technical writing 의 hierarchy

README: 매 30-second pitch.
Tutorial: 매 narrative, hand-holding.
How-to: 매 task-oriented.
Reference: 매 exhaustive.
Explanation: 매 conceptual.

→ Diátaxis framework.

Anti-clarity 의 source

매 jargon 의 over-use.
매 passive voice.
매 nominalization ("perform an analysis" > "analyze").
매 abstract noun ("optimization", "leverage").
매 throat-clearing ("It is important to note that...").

💻 패턴

Prompt template (specific + constraint)

Role: Senior TypeScript backend engineer.
Task: Refactor this function for testability.
Constraints:
- Keep the public signature unchanged.
- Extract DB call to a repository interface.
- Return a Result<T, E> instead of throwing.
- Output: code only, no explanation.

Input:
{{code}}

→ 매 vague "make it better" 의 X.

PR description template

## Why
[1-2 sentence motivation]

## What changed
- bullet
- bullet

## How to verify
- [ ] step 1
- [ ] step 2

## Risk
[regression area / rollback plan]

## Out of scope
[what NOT done — prevent reviewer churn]

Plain language rewrite

❌ Original (50 words):
"It is recommended that users should consider implementing
appropriate validation mechanisms in order to ensure that 
input data is properly sanitized before being processed."

✅ Rewrite (10 words):
"Validate input before processing."

Articulateness 의 self-check

1. 매 첫 sentence 의 conclusion?
2. 매 sentence avg < 25 word?
3. 매 active voice 의 80%+?
4. 매 jargon 의 explain or replace?
5. 매 reader 의 next step 의 clear?

🤔 결정 기준

상황	우선
Spec / RFC	Structural clarity + precision
README	BLUF + plain language
PR description	Why + what + risk
LLM prompt	Specific + constraint + format
Slack / async	Concise + actionable
Customer-facing	Plain + nuance

기본값: BLUF + active + concise + concrete.

🔗 Graph

🤖 LLM 활용

언제: 매 prompt 의 craft. 매 doc / spec / PR write. 매 communication 의 review. 언제 X: 매 creative writing 의 nuance 의 sacrifice. 매 poetic context.

❌ 안티패턴

Throat-clearing: 매 "It is important to note that..."
Passive voice abuse: 매 actor 의 hide.
Jargon dump: 매 audience 의 ignore.
Burying lede: 매 conclusion 의 끝.
Nominalization: "perform an analysis" 의 "analyze".
Vague prompt: 매 "make it better" 의 LLM 에 unhelpful.

🧪 검증 / 중복

Verified (Plain Writing Act, GOV.UK style, Diátaxis).
신뢰도 B.
Related: Prompt_Engineering · Technical-Writing · Plain-Language.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — BLUF + Plain Language + Diátaxis + prompt template

4.9 KiB Raw Blame History