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

4.8 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

Team Topologies

매 한 줄

"매 4 fundamental team types + 3 interaction modes 의 fast flow 의 organize". Matthew Skelton & Manuel Pais (2019) 의 framework — Conway's Law 의 deliberately inverse-leverage. 2026 모던 SaaS scale-up 매 default playbook.

매 핵심

매 4 Team Types

Stream-Aligned: 매 single value stream (product/feature/customer) 의 own. 매 most teams (~70%).
Platform: 매 internal services (CI/CD, observability, auth) 의 stream-aligned 의 enable.
Enabling: 매 short-term coaching (e.g. "help adopt OpenTelemetry"). 매 disband after.
Complicated-Subsystem: 매 deep specialist domain (video codec, ML inference, payments crypto).

매 3 Interaction Modes

Collaboration: 매 high-bandwidth, short-term, exploratory.
X-as-a-Service: 매 platform team의 well-defined API 의 provide.
Facilitating: 매 enabling team의 coach mode.

매 Cognitive Load

매 team의 cognitive load 의 limit (Miller's 7±2). 매 boundaries 의 set.
매 intrinsic / extraneous / germane load 의 distinguish.

매 응용

Scale-up 50→500 eng — stream-aligned squad 의 split.
Platform team 의 internal-developer-platform (IDP) build.
ML platform — 매 complicated-subsystem (training infra) + platform (serving).

💻 패턴

Team API (markdown contract)

# Team API: Payments Platform

## Mission
Provide reliable payment processing API for stream-aligned teams.

## Services Provided (X-as-a-Service)
- POST /charge (SLO 99.95%)
- POST /refund (SLO 99.9%)

## On-call
PagerDuty: payments-platform-oncall

## Interaction
- X-as-a-Service for stream-aligned teams.
- Collaboration window: Tuesdays 10-11am for new integrations.

Stream-aligned team boundary

# team-checkout.yaml
team: checkout-squad
type: stream-aligned
owns:
  - service: checkout-api
  - service: cart-service
  - frontend: /checkout/*
depends_on:
  - team: payments-platform (X-as-a-Service)
  - team: identity-platform (X-as-a-Service)
oncall: checkout-oncall

Cognitive load assessment

# Quick survey, 1-5 scale per team
load_survey = {
    "domain_complexity": 4,   # how complex is the business?
    "tech_complexity": 3,     # how many techs to master?
    "context_switches": 5,    # how many systems do you touch?
    "external_deps": 2,       # how many other teams must you coordinate with?
}
score = sum(load_survey.values())
# >15: overloaded, consider splitting or moving deps to platform

Enabling team engagement

# Engagement: Observability Adoption
Enabling team: SRE-Coaching
Target: checkout-squad
Duration: 6 weeks
Goal: Adopt OpenTelemetry tracing, define SLOs.
Exit criteria: Team independently maintains SLO dashboard.

매 결정 기준

상황	Team type
매 customer-facing product slice	Stream-aligned
매 shared infra (k8s, CI/CD)	Platform
매 short-term capability gap	Enabling
매 deep specialist (codec, ML kernel)	Complicated-Subsystem
매 ad-hoc cross-team feature	Collaboration mode (temporary)

기본값: 매 stream-aligned 의 default. 매 platform team의 too-early formation 의 avoid (먼저 stream-aligned 의 pain point 의 ).

🔗 Graph

부모: Conway's Law · DevOps
변형: Spotify Model
응용: Platform Engineering · Internal Developer Platform
Adjacent: Domain-Driven Design · Microservices · SRE

🤖 LLM 활용

언제: 매 50+ engineer org 의 redesign. 매 platform team의 charter 의 draft. 매 cognitive load survey 의 analyze. 언제 X: 매 <10 person startup (premature). 매 Conway 의 ignored 매 consulting deliverable.

❌ 안티패턴

Platform-first: 매 stream-aligned 매 pain 없이 platform 의 build → unused.
Permanent enabling team: 매 coaching team의 forever 의 stay → "ivory tower".
Component team: 매 horizontal slice (e.g. "frontend team") — 매 stream 의 cut, hand-offs ↑.
Too many interactions: 매 every team의 every team 의 talk → 매 N² coordination cost.

🧪 검증 / 중복

Verified (Skelton & Pais, "Team Topologies" 2019; teamtopologies.com 2026 case studies).
신뢰도 A.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — Team Topologies 4-type + interaction modes + Team API

4.8 KiB Raw Blame History