agentleFS
idoforgodidoforgod/cys-bookwriting-skills/skills/bookwriting-absorb-master/SKILL.md

bookwriting-absorb-master

Skill4 starsChanged 3 months agoMITView on GitHub

저자 책 ① 단계 마스터 — 진입 정밀도. inbox/import 자산을 4 type (seed·evidence·insight·cycle-input)으로 분류·시드 등록·라우팅 제안. frontmatter 우선 + 자동 추측 fallback. M7 IdGenerator 결정론 호출. 옵션 A+ 자기완결.

The file

Files
SKILL.md

bookwriting-absorb-master (① 진입 마스터)

TLDR

저자 외부 manual 호출 결과 (foresight 사슬·NotebookLM bridge·시드 입력)를 진입 정밀도로 분류·등록. 모든 후속 단계의 입력 품질이 ① absorb의 정밀도로 결정.

Triggers

  • /bookwriting cycle <target> Step 2 자동
  • /bookwriting absorb 또는 /bookwriting inbox process — inbox 일괄 흡수
  • /bookwriting import <path> --as <type> [--to|--link] — 즉시 import
  • /bookwriting-absorb classify <path> — 분류만

Detailed Methodology

1. 4 type 자동 분류 (Q13-D-2)

type트리거 키워드·패턴라우팅
seed짧은 명제·통찰·1문장seed-registry
evidence출처·인용·통계 풍부chapters/<ch>/evidence-track
insight"통찰·결론·해석" 단어chapter pending-insights
cycle-input외골격·구조·flowbookwriting-master 직접 전달

frontmatter type 명시 시 → 즉시 사용 frontmatter 없음 → 자동 추측 (confidence 점수) → < 0.7 시 저자 묶음

2. 라우팅 cascade L1·L2·L3 (Q20-D-3)

import paths
from relevance_scorer import RelevanceScorer
from pathlib import Path

# L1: skeleton.md 키워드 매칭
skeleton = paths.skeleton_path()
scorer = RelevanceScorer([skeleton])
score_l1 = scorer.score(seed_text, skeleton)

if score_l1 >= 0.8:
    # 명확 → 라우팅 결정
    ...
elif score_l1 < 0.8:
    # L2 Claude 의미 평가 → L3 Obsidian 풀
    ...

3. Citation Gate Medium만 적용 (예외)

저자 흡수 결정 자유 보장 — Critical 즉시 차단 유일한 예외. citation-gate.md §3 Philosophy 명시. 후속 ② fact-anchor 이후로는 Critical 즉시 차단.

4. ID 발급 (M7 결정론)

import paths
from id_generator import IdGenerator
import yaml
from pathlib import Path

registry_path = paths.seed_registry_path()
with open(registry_path) as f:
    registry = yaml.safe_load(f.read())

current_ids = [s.get("id") for s in (registry.get("seeds") or [])]
max_num = IdGenerator.compute_max_num(current_ids, "S")

gen = IdGenerator(registry_path)
new_seed_id = gen.issue("seed", current_max_num=max_num)

5. inbox 정리 (atomic write)

from atomic_writer import AtomicWriter
import shutil

# 처리 완료 → _processed/ 이동
processed_dir = inbox_dir / "_processed"
processed_dir.mkdir(exist_ok=True)
shutil.move(str(inbox_file), str(processed_dir / inbox_file.name))

저자 진북 정합

진북검증
#1 할루시네이션 0✅ M2 Medium 감지 + 흡수 자유 명시 예외
#3 출처 명시✅ frontmatter source·citation 메타 추출
#4 저자 작가성✅ 저자 명시 호출 우선 (--as·--to)
#7 결정론 환원✅ M7 IdGenerator + M6 RelevanceScorer 결정론 cascade

출처

SPEC: specs/bookwriting-absorb-master.md (17 항목 풀) 실행 모듈: lib/id_generator.py + lib/relevance_scorer.py + lib/citation_detection.py 호출자: bookwriting-master Step 2

Sub-skill Orchestration Trace ⭐ (저자 7번째 절대 protocol)

본 마스터 호출 시 4개 sub-skill을 자동 orchestration. M18 OrchestratorTracer가 각 호출을 결정론 기록.

순서Sub-skill호출 시점AI Agent 페르소나결정론 모듈
1bookwriting-absorb-frontmatter-parserStep 2Frontmatter Parseryaml.safe_load
2bookwriting-absorb-type-classifierStep 3 (frontmatter type 누락 시)Type Classifier규칙 매칭 + LLM eval fallback
3bookwriting-absorb-route-cascade ⭐Step 6Route CascadeM6 RelevanceScorer (BM25)
4bookwriting-absorb-inbox-organizerStep 9Inbox Organizerpathlib·shutil

자동 호출 흐름

사용자 명령: /bookwriting-absorb <mode> [target]
↓
Step 2: bookwriting-absorb-frontmatter-parser 자동 호출
  Input: inbox/*.md 또는 --target <path>
  Output: frontmatter dict + 누락 필드 목록
↓ (type 누락 시)
Step 3: bookwriting-absorb-type-classifier 자동 호출
  Input: 본문 + partial frontmatter
  Output: TypeGuess (type_value, confidence, method)
  ⚠ confidence < 0.7 → 저자 묶음 escalate
↓ (--to·--link 명시 없을 시)
Step 6: bookwriting-absorb-route-cascade 자동 호출
  Input: type 분류된 자산 + book-config search_paths
  Output: RouteCandidate (chapter, layer, section, cascade_layer L1/L2/L3)
↓
Step 9: bookwriting-absorb-inbox-organizer 자동 호출
  Input: 처리 결과 list[dict]
  Output: _processed/·_failed/ 이동 + 통계
↓
Master Synthesis: cycle-registry.outputs 갱신 + 저자 화면 출력

저자 결정 묶음 강제 (자동 결정 영구 금지)

  • Type Classifier confidence < 0.7 → 저자 묶음
  • Route Cascade L3까지도 confidence < 0.5 → unrouted + 저자 묶음

OrchestratorTracer (M18) 자동 기록 양식

from orchestrator_tracer import OrchestratorTracer
tracer = OrchestratorTracer(master_name="bookwriting-absorb-master")
tracer.record(skill_name="bookwriting-absorb-frontmatter-parser", sequence=1, ...)
tracer.record(skill_name="bookwriting-absorb-type-classifier", sequence=2, ...)
tracer.record(skill_name="bookwriting-absorb-route-cascade", sequence=3, ...)
tracer.record(skill_name="bookwriting-absorb-inbox-organizer", sequence=4, ...)
cycle_entry["orchestrator_trace"] = tracer.export()

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.