القائمة
개인 어휘 그래프 — 한↔영 크로스링귀얼 쿼리 확장 + 한국어 형태소 분석. expand로 한글 단어에서 영어 태그 후보를 찾고, stem으로 한국어 어간을 추출한다. knowledge_search의 3층 역할. '단어 확장', '태그 찾기', '한영 매핑', 'expand', 'stem', '형태소'.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Use tmux instead of bash tool to run commands that take more than ~30 seconds, like bulk operations, db migrations, dev servers.
Google Workspace + Search Console all-in-one CLI (gog). Calendar, Gmail, Drive, Tasks, Chat, Contacts, Sheets, Docs, Search Console, YouTube, Photos, Meet, and more. Single binary. Source: steipete/gogcli (upstream).
GLG의 시간축 관측소 (범용 스킬) — 어느 세션·리포에서든 GLG의 하루를 깊이 0/1/2/3(시간블록·저널헤딩·에이전트스탬프·커밋/노트)으로 한 KST 축 위에 세운다. scripts/collect.py로 LOCAL FULL(events.jsonl, 로컬·gitignore)을 만들고 query.py로 꺼낸다. 스크립트는 HOME 앵커(~/repos·~/org·lifetract·agenda)라 CWD와 무관하게 돈다. 이 스킬을 읽지 않고 축을 건드리면 계약을 다시 도출하다 틀린다 — 고정 KST, 반개방 [from,to), 자정은 시작일 귀속, 0건은 ok가 아니다, sqlite를 직접 열지 않는다, 코멘트는 어떤 projection에도 나가지 않는다. Triggers: timeline, 관측소, 시간축, 깊이 0, depth axis, collect.py, query.py, events.jsonl, LOCAL FULL, 하루 slice, 주간 판독, 도메인 분포, projection, 표현, 증거, provenance, 커밋 수, 저널 일수, 사실 확인.
Incrementally embed sessions only — near-live. On call, new sessions land in semantic-memory immediately. OpenRouter Qwen3-Embedding-8B 4096d, paid remote but ~$0.000–0.001 for a few recent sessions. dim 4096 preflight → API-0 exit when to_index=0. Use before a new session or when recent-session recall feels stale. '/memory-sync', 'memory sync', 'session embedding', '세션 임베딩', '세션 증분', '기억 최신화'.
Semantic search over past sessions (pi + Claude Code) and the public digital garden md index (andenken md.lance, OpenRouter Qwen3-Embedding-8B 4096d). Uses LanceDB + hybrid retrieval (vector + FTS with score normalization). Korean↔English cross-lingual via dictcli expand. Recall tracking for memory consolidation. Use when searching for past conversations, decisions, context, or garden knowledge concepts.
Emacs daemon — org manipulation, denote notes, citar bibliography, org-agenda, arbitrary elisp. Two sockets: server (agent work), user (show file to user). Core: agent-denote-add-history(ID,CONTENT), agent-denote-add-heading(ID,TITLE,BODY) or (ID,TITLE,TAG,BODY) — no tag? body as 3rd arg. Never pass nil. agent-denote-add-link(ID,TARGET-ID,DESC). All 3 args required.
| name | dictcli |
| description | 개인 어휘 그래프 — 한↔영 크로스링귀얼 쿼리 확장 + 한국어 형태소 분석. expand로 한글 단어에서 영어 태그 후보를 찾고, stem으로 한국어 어간을 추출한다. knowledge_search의 3층 역할. '단어 확장', '태그 찾기', '한영 매핑', 'expand', 'stem', '형태소'. |
Korean↔English triple graph. Expands Korean words to English Denote tags. Layer 3 of knowledge_search — cross-lingual query enrichment.
Binary bundled in skill directory. Must cd to {baseDir} (reads CWD/graph.edn).
| Command | Usage | Output |
|---|---|---|
| expand | cd {baseDir} && ./dictcli expand "보편" --json | ["universal","universalism","particular","paideia"] |
| graph | cd {baseDir} && ./dictcli graph "보편" | All triples for the word |
| stats | cd {baseDir} && ./dictcli stats | Graph statistics |
| validate | cd {baseDir} && ./dictcli validate | Invariant check |
Traverses the graph: direct translation → opposite's translation → related word's translation.
cd {baseDir} && ./dictcli expand "기술" --json # → ["art","technology","technique"]
cd {baseDir} && ./dictcli expand "도피" --json # → ["escape","flight","avoidance","evasion"]
Empty array = word not in graph.edn. Fall back to original query.
1. User: "보편 학문 관련 노트"
2. expand "보편" → ["universal","universalism","paideia"...]
3. Search: "보편 학문 universal universalism paideia"
Agent autonomously: extract concept words → expand → enrich query → knowledge_search.
JVM-only (Kiwi JNI — not in native binary). Called via dictcli repo's run.sh stem.
Agents rarely call directly — andenken batch-indexes via --batch / --serve mode.
cd ~/repos/gh/dictcli && ./run.sh stem "설계했다" # → 설계
cd ~/repos/gh/dictcli && ./run.sh stem --serve 9876 # socket server (1ms/query)
Pipeline: "설계했다" → stem → "설계" → expand → ["design","architecture"]
graph.edn: 3,971 triples, 2,449 :trans mappings, 4,728 words, 526 clusters:trans(2449), :source(1422), :related(44), :synonym(44)| Component | Runtime | Latency |
|---|---|---|
| expand/lookup/stats | GraalVM native-image | ~9ms |
| stem | JVM (Kiwi JNI) | ~1ms (server) / ~3s (cold start) |
Native binary cannot include stem (JNI incompatible with native-image). Two separate execution paths — no conflict.
| Layer | Tool | Role |
|---|---|---|
| 1 | knowledge_search | Embedding vector search |
| 2 | denotecli | Exact match + graph links |
| 3 | dictcli expand + stem | Korean→English query expansion + stemming |