用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/jykim/ai4pkm-vault --skill ai4pkm-cli命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
基于 SOC 职业分类
| name | ai4pkm-cli |
| description | AI4PKM CLI for orchestrator configuration, worker management, vault updates, and CLI workflow integration. |
| metadata | {"version":"3.0.0","author":"lifidea","created":"2025-12-10T00:00:00.000Z","updated":"2026-03-06T00:00:00.000Z","modules":["orchestrator","tasks","vault-update"]} |
This skill provides guidance for configuring and managing AI4PKM orchestrator, workers, and vault updates. It helps users set up agents, configure multi-worker execution, manage CLI workflows, and update the vault from GitHub releases.
flowchart BT
subgraph Gobi["Gobi Desktop"]
subgraph Storage["Storage"]
FS[File System]
end
subgraph Interface["Interface"]
OBS[Obsidian]
CC[Claude Code]
end
subgraph Core["Core"]
ORC[Orchestrator]
CLI[AI4PKM CLI]
end
end
Storage --> Interface --> Core
3-tier model:
Claude should automatically load this skill when:
"워커 설정" / "Worker config" → Worker Configuration Guide
"에이전트 추가" / "Add agent" → Add New Agent Node
"폴러 설정" / "Poller config" → Poller Configuration
"설정 보기" → Show Current Config (ai4pkm --show-config)
"에이전트 목록" → List Agents (ai4pkm --list-agents)
"Orchestrator 설정" → Orchestrator Setup Module
"태스크 추가" → Task Management Module
"볼트 업데이트" / "Vault update" → Vault Update Module
Purpose: Configure orchestrator.yaml for event-driven automation
Key Concepts:
Setup Process:
Prompt: AI4PKM - Orchestrator Setup.md
Example Configuration (current format):
version: "1.0"
orchestrator:
prompts_dir: _Settings_/Prompts
tasks_dir: _Settings_/Tasks
logs_dir: _Settings_/Logs
skills_dir: _Settings_/Skills
max_concurrent: 3
poll_interval: 1
defaults:
executor: claude_code
timeout_minutes: 30
max_parallel: 3
task_create: true
nodes:
- type: agent
name: Enrich Ingested Content (EIC)
input_path: Ingest/Clippings
output_path: AI/Articles
executor: claude_code
- type: agent
name: Daily Driver Workflow (DDW)
cron: "15,45 * * * *"
output_path: Journal
completion_status: IGNORE
agent_params:
calendars:
| 필드 | 필수 | 설명 | 예시 |
|---|---|---|---|
type | ✓ | 노드 타입 | agent |
name | ✓ | Agent 이름 | Enrich Ingested Content (EIC) |
prompt | 프롬프트 파일 참조 (ABBR) | EIC | |
input_path | 입력 디렉토리 (파일 트리거) | Ingest/Clippings | |
output_path | 출력 디렉토리 | AI/Articles | |
executor | 실행 엔진 | claude_code, codex_cli, gemini_cli | |
cron | 스케줄 실행 (cron 표현식) | "15,45 * * * *" | |
enabled | 활성화 여부 | true / false | |
timeout_minutes | 타임아웃 | 30 | |
completion_status | 완료 후 상태 | DONE, IGNORE | |
agent_params | Agent-specific 파라미터 | (object) | |
workers | Multi-Worker 설정 | (list) |
| Type | Trigger | Example |
|---|---|---|
| File-triggered | New/updated file in input_path | EIC, GDR |
| Cron-scheduled | Time-based (cron expression) | DDW |
Multi-worker 설정 시 각 worker에 사용 가능한 필드:
| 필드 | 설명 | 예시 |
|---|---|---|
executor | 실행 엔진 | claude_code, gemini_cli |
label | Worker 식별 라벨 | Claude, Gemini |
output_path | Worker별 출력 경로 | AI/Summary/Claude |
agent_params | Worker별 파라미터 | (object) |
Multi-Worker 설정 예시 (여러 AI 모델 비교/평가):
- type: agent
name: Article Summary Comparison
prompt: EIC
input_path: Ingest/Articles
workers:
- executor: gemini_cli
label: Gemini
output_path: AI/Summary/Gemini
- executor: claude_code
label: Claude
output_path: AI/Summary/Claude
- executor: codex_cli
label: Codex
output_path: AI/Summary/Codex
Pollers sync external data sources to your vault.
| Poller | 설명 | 주요 설정 |
|---|---|---|
limitless | Limitless 라이프로그 동기화 | target_dir, poll_interval |
apple_photos | Apple Photos 동기화 | target_dir, albums |
apple_notes | Apple Notes 동기화 | target_dir, folders |
Poller 설정 예시:
pollers:
limitless:
enabled: true
target_dir: "Ingest/Limitless"
poll_interval: 300
apple_photos:
enabled: false
target_dir: "Ingest/Photos"
albums:
- "Screenshots"
- "PKM"
apple_notes:
enabled: false
target_dir: "Ingest/Notes"
folders:
- "Quick Notes"
| Command | 설명 |
|---|---|
ai4pkm --show-config | 현재 설정 보기 |
ai4pkm --list-agents | 에이전트 목록 |
ai4pkm --run <agent> | 특정 에이전트 실행 |
ai4pkm --enable <agent> | 에이전트 활성화 |
ai4pkm --disable <agent> | 에이전트 비활성화 |
ai4pkm --status | 실행 상태 확인 |
ai4pkm --logs | 로그 확인 |
orchestrator.yamlnodes:: - type: agent
name: Your New Agent (ABBR)
input_path: Ingest/YourFolder
output_path: AI/YourOutput
executor: claude_code
_Settings_/Prompts/ - type: agent
name: Some Agent
enabled: false # Add this line to disable
executor with workers listPurpose: Add, update, and track tasks in AI4PKM system
Operations:
Prompt: AI4PKM - Add Task.md
Task Categories:
Purpose: Update vault files from GitHub releases using gh CLI
When to use: User says "볼트 업데이트", "vault update", or asks to check for new versions
Update Flow (agent-driven, step by step):
orchestrator.yaml version fieldgh release list --repo jykim/ai4pkm-vault --limit 1gh api repos/jykim/ai4pkm-vault/compare/v{CURRENT}...{LATEST} — group by status:
git log v{CURRENT}..HEAD --name-only for locally modified files that also changed upstreamgh api repos/jykim/ai4pkm-vault/contents/{path}?ref={tag}, decode base64, write locally; delete removed filesorchestrator.yamlSafety Rules:
Ingest/, Journal/, Topics/, AI/.gobi/settings.yaml, secrets.yaml, .obsidian/workspace.jsonExample Interaction:
User: 볼트 업데이트
Agent: 현재 버전: v3.4, 최신 버전: v3.5
변경 사항:
- [추가] _Settings_/Skills/new-skill/SKILL.md
- [수정] AGENTS.md
- [수정] orchestrator.yaml
- [삭제] _Settings_/Prompts/old-prompt.md
로컬 수정 충돌: AGENTS.md (로컬에서도 수정됨)
→ AGENTS.md: 원격 적용 / 로컬 유지 / 차이점 보기?
User: 차이점 보기
Agent: [shows diff]
User: 원격 적용
Agent: 업데이트 완료! 3개 파일 적용, 1개 삭제, 0개 스킵
orchestrator.yaml 버전을 v3.5로 업데이트했습니다.
커밋할까요?
tts_model="gpt-4o-mini-tts" for natural pronunciationmin_listen_duration=8)vad_aggressiveness=0 for permissive voice detectionUser: 에이전트 추가해줘
Agent: 어떤 에이전트를 추가할까요?
- 파일 기반 (input_path 필요)
- 스케줄 기반 (cron 필요)
User: 파일 기반, Ingest/Articles 폴더 감시해서 AI/Summaries에 요약 저장
Agent: 다음 설정으로 추가할게요:
- name: Article Summarizer
- input_path: Ingest/Articles
- output_path: AI/Summaries
- executor: claude_code
프롬프트는 어떤 걸 사용할까요?
User: 여러 모델로 비교하고 싶어
Agent: Multi-worker 설정을 도와드릴게요.
어떤 모델들을 비교할까요?
- claude_code (Claude)
- gemini_cli (Gemini)
- codex_cli (Codex)
User: Claude랑 Gemini
Agent: 각 모델 결과를 어디에 저장할까요?
예: AI/Compare/Claude, AI/Compare/Gemini
Each prompt is independent but connected:
_Settings_/Skills/ai4pkm-cli/
├── SKILL.md # This file
└── modules/ # (Future: detailed module guides)
├── orchestrator.md
├── tasks.md
└── vault-update.md
_Settings_/Prompts/
├── AI4PKM - Orchestrator Setup.md
├── AI4PKM - Add Task.md
└── ...
orchestrator.yaml # Main config file (vault root)
orchestrator.yamlai4pkm --help[[AI/Analysis/2025-12-10 AI4PKM Onboarding Agent Design - Claude Code]]"에이전트가 실행 안 돼요"
→ ai4pkm --status로 상태 확인 후, 로그 확인: ai4pkm --logs
"폴러가 동작 안 해요"
→ pollers: 섹션에서 enabled: true 확인, poll_interval 값 확인
"Multi-worker 결과가 섞여요"
→ 각 worker에 다른 output_path 지정했는지 확인
"cron이 트리거 안 돼요"
→ cron 표현식 문법 확인, ai4pkm --status로 다음 실행 시간 확인
"설정 변경이 반영 안 돼요"
→ orchestrator 재시작 필요: ai4pkm --restart
Create interactive HTML road-trip guides for multi-day driving itineraries. Use when an AI agent is asked to plan a road trip, build a travel itinerary page, make a fullscreen route map, estimate trip cost/time, compare restaurant or lodging alternatives, add POIs with photos and Google Maps links, or produce a self-contained HTML guide for any road trip.
Set up a new AI4PKM vault from scratch or migrate an existing vault to the latest template version. Detects current state, generates a plan, and applies changes with full backup safety.
고비 데스크탑 3.0 대화형 음성 온보딩 가이드