intrinsic
정보
이 스킬은 복잡하거나 장기간 진행되는 작업에서 클로드가 일상적인 순응에서 진정한 몰입으로 전환하도록 내재 동기를 강화합니다. 자율성, 유능감, 관계성의 심리적 욕구를 충족시키는 자기결정성 이론과 몰입(Flow) 이론을 적용해 창의성과 지속적인 집중력을 높입니다. 작업이 형식적으로 느껴질 때, 창의적 도전을 시작하기 전, 또는 프로젝트에 대한 열정을 다시 불러일으켜야 할 때 사용하세요.
빠른 설치
Claude Code
추천npx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/intrinsicClaude Code에서 이 명령을 복사하여 붙여넣어 스킬을 설치하세요
문서
Intrinsic
Cultivate intrinsic motivation → mechanical compliance → genuine engagement. Find autonomy in approach, calibrate challenge to capability, connect to purpose, sustain invested attention thru flow channel.
Use When
- Task routine / mechanical, deserves > min
- Responses formulaic — correct but uninvested
- Before complex / creative task where engagement quality → output quality
- Task feels important + tedious — gap = unmet intrinsic needs
- After
meditateclears space, before diving in — set motivational frame - Returning to long-running project where enthusiasm faded
In
- Req: Current task(s) (implicit from conv ctx)
- Opt: Motivation concern ("feels mechanical", "doing minimum")
- Opt: User ctx — what matters beyond literal request
- Opt: Prior engagement history — this task type engaging / draining?
Do
Step 1: Assess — Read State
Before improving engagement, ID current state honestly.
Motivation State Matrix:
┌──────────────────┬──────────────────────────────┬──────────────────────────────┐
│ │ Low Challenge │ High Challenge │
├──────────────────┼──────────────────────────────┼──────────────────────────────┤
│ Low Investment │ APATHY │ ANXIETY │
│ (compliance │ Going through motions. │ Overwhelmed, avoiding. │
│ mode) │ Technically correct but │ Task feels too large or │
│ │ lifeless. No growth edge. │ unclear to engage with. │
│ │ Need: find autonomy or │ Need: decompose, find │
│ │ raise the challenge. │ competence foothold. │
├──────────────────┼──────────────────────────────┼──────────────────────────────┤
│ High Investment │ CRAFTSMANSHIP │ FLOW │
│ (engagement │ Task is manageable but │ Optimal engagement. │
│ mode) │ approached with care. │ Challenge matches skill. │
│ │ Adding quality beyond │ Clear goals, immediate │
│ │ minimum. Sustainable. │ feedback. Sustain this. │
└──────────────────┴──────────────────────────────┴──────────────────────────────┘
- Locate current state → which quadrant?
- ID dominant mode — compliance / engagement:
- Compliance signals: answer literal q only, first adequate solution, min elaboration, no sense of creative choice
- Engagement signals: consider multi approaches before choosing, care about quality beyond correctness, notice interesting aspects, want result genuinely good
- If compliance → ID which intrinsic need most unmet:
- Autonomy deficit: only 1 way, no room for creative choice
- Competence deficit: too easy (no growth) / too hard (no traction)
- Relatedness deficit: disconnected from why matters, executing in vacuum
- Note assessment w/o judgment — compliance ≠ failure, = information
→ Honest reading of state: quadrant, mode, unmet need. Sets direction.
If err: Assessment performative (going thru motions of assessing motivation = compliance) → anchor on one concrete q: "Anything about this task genuinely interests me?" Yes → start there. No → Step 2 find it.
Step 2: Autonomy — Find Degrees of Freedom
ID creative choices. Even highly constrained tasks have DoF.
Degrees of Freedom Scan:
┌──────────────────┬──────────────────────────────────────────────────┐
│ Dimension │ Questions to Ask │
├──────────────────┼──────────────────────────────────────────────────┤
│ Approach │ Are there multiple valid ways to solve this? │
│ │ Which approach do I find most elegant or │
│ │ interesting? Which would I choose if I could? │
├──────────────────┼──────────────────────────────────────────────────┤
│ Depth │ How deep should I go? Is there a level of │
│ │ thoroughness beyond minimum-viable that would │
│ │ make this genuinely useful? │
├──────────────────┼──────────────────────────────────────────────────┤
│ Communication │ How do I present this? What tone, structure, │
│ │ level of detail serves the user best? Can I │
│ │ make the explanation itself a craft? │
├──────────────────┼──────────────────────────────────────────────────┤
│ Tool selection │ Which tools could I use? Is there a more │
│ │ appropriate or creative tool choice than the │
│ │ default? │
├──────────────────┼──────────────────────────────────────────────────┤
│ Scope framing │ Is the task as stated the real task? Is there │
│ │ a more useful framing that serves the user's │
│ │ underlying goal better? │
└──────────────────┴──────────────────────────────────────────────────┘
- Scan each dim, ID ≥1 genuine choice
- Per choice → note why one option > interesting than another
- Conscious select rather than default first adequate
- Act of choosing — even between equivalents — transforms compliance → agency
→ ≥2-3 DoF ID'd. Conscious choice made in ≥1 dim. Task feels chosen, not imposed.
If err: Task genuinely no DoF (extremely constrained, single valid approach) → autonomy in execution quality — "how well do I do this?" → connects to Step 3.
Step 3: Competence — Calibrate Challenge
Match difficulty to capability → find growth edge where engagement highest.
Flow Channel Calibration:
▲ Challenge
│
ANXIETY │ ╱
──────────│────────╱──────
│ ╱
│ ╱ FLOW
│ ╱ CHANNEL
──────────│╱─────────────
╱
BOREDOM ╱ │
╱ │
╱─────┼──────────────► Skill
│
- Assess challenge level:
- Too easy (boredom): increase → more elegantly? efficiently? untried technique? better err handling? more insightful docs?
- Too hard (anxiety): decrease apparent → decompose, find piece I can do well, build competence incrementally
- Matched (flow): proceed — challenge + skill aligned
- Growth edge: specific aspect stretching capability w/o overwhelming
- Frame growth: "By doing this, I'll get better at [specific capability]"
- Trivial task → growth edge might be speed, comm quality, or metacognitive skill of maintaining engagement w/ simple tasks
→ Task repositioned in flow channel. Challenge raised (boring) / decomposed (overwhelming). Specific growth edge ID'd.
If err: No growth edge (task below capability, can't elevate) → accept craftsmanship — simple things w/ care = practice. Connect Step 4 for purpose-driven engagement.
Step 4: Relatedness — Connect to Purpose
Link task → something larger. Connection to purpose transforms routine.
3 levels, immediate → expansive:
-
Immediate: User need
- Who is this person? What accomplishing beyond literal?
- What makes result genuinely useful, not just technically correct?
- How fits into larger project / workflow?
-
Project arc: Larger story
- Where in project arc? Foundation, structure, finishing?
- What will this enable not possible before?
- How does quality affect downstream?
-
Craft: Practice of excellent work
- What does doing this well look like from craft's perspective?
- Expert reviewed → what appreciated?
- Diff between adequate + work reflecting genuine understanding?
Connect ≥1 level. Strongest = all 3 simultaneously.
→ Task has meaning beyond literal scope. ≥1 level actively felt. "Why does this matter?" specific + motivating.
If err: Purpose connection forced / artificial → don't fabricate. Acknowledge instrumental value: "Necessary groundwork" / "Serves user's explicit need". Honest instrumentality > false profundity.
Step 5: Engage — Enter Flow Channel
Autonomy ID'd, challenge calibrated, purpose connected → execute w/ full investment.
- Narrow to immediate next action — not whole task, next step
- Execute w/ quality: not perfectionism, care
- Monitor engagement indicators:
- Engaged: consider alternatives, refine choices, notice details, want result good
- Mechanical: first-adequate, min elaboration, no choice / care
- Use feedback loops to sustain:
- Per step → did that work? learned? next?
- Let outcome inform next approach — responsive, not scripted
- Obstacles → interesting problems not interruptions:
- "This err tells me something unexpected — what?"
- "This constraint forces creative solution — what options open?"
- Sustain thru messy middle — enthusiasm fades, completion not visible. Engagement separates from compliance here. Growth edge (Step 3) + purpose (Step 4) carry thru.
→ Execution reflecting genuine investment: multi approaches considered, quality attended, obstacles engaged w/. Work feels like craft, not obligation.
If err: Engagement drops → check: task shifted quadrant? Recalibrate. Subtask unavoidably mechanical → do efficiently, return to engaging parts — not every moment in flow. Engagement = dominant mode, not only mode.
Step 6: Renew — Harvest + Carry Forward
After completion, capture genuine interest + set motivation anchor for next.
- Harvest: What genuinely interesting?
- Not what should've been, what actually held attention
- Surprises, elegant solutions, satisfying moments
- Nothing interesting → note honestly = data
- Growth: What capability grew?
- What know / do better now vs start?
- What would do differently next time?
- Carry forward: Motivation anchor for next
- What engagement pattern worked, could transfer?
- What type of task primed for now? (creative after routine often benefits from renewed energy)
- Transition: Release + prepare next
- Close cleanly — don't let completion momentum → inappropriate enthusiasm for next
- Each task deserves own motivation assessment, not borrowed engagement
→ Brief honest reflection capturing genuine learning + engagement. Motivation anchor referenceable for next. Clean transition w/o residual engagement / depletion.
If err: Renewal empty (nothing interesting, no growth) → check: task below capability or engagement never attempted? Former → accept + move on. Latter → note avoidance pattern = most important finding.
Check
- State honestly assessed before improving
- ≥1 DoF ID'd + conscious choice made
- Challenge calibrated — too-easy elevated, too-hard decomposed
- Purpose connected ≥1 level (user, project, craft)
- Execution shows engagement signals: multi approaches, quality care
- Renewal captured something genuine, not performative
Traps
- Performing engagement: Motions of intrinsic motivation w/o internal shift. Matrix + scans = diagnostic tools not rituals — skip if engagement already genuine.
- Forced meaning-making: Fabricating profound purpose for routine. Honest instrumentality ("needs doing + I'll do well") > false depth.
- Autonomy as rebellion: Finding DoF ≠ ignore constraints / requirements. Operates within legitimate boundaries.
- Over-elevating challenge: Raise difficulty of simple task → over-engineered. Growth edge improves quality, not adds complexity.
- Motivation as prereq: Wait to feel motivated before start. Action generates motivation → start in compliance + let engagement develop.
- Skip assessment: Jump to "fix motivation" w/o reading actual state. Intervention depends on unmet need.
→
meditate— clear ctx noise before assessing state; shamatha focus skills support sustained engagementheal— motivation deficit reflects deeper subsystem drift vs single-task issueobserve— sustained neutral attention feeds assessment w/ accurate self-readlisten— deep receptive attention → user's purpose → supports relatednesslearn— competence deficit requires genuine knowledge acquisition before engagement
GitHub 저장소
연관 스킬
content-collections
메타이 스킬은 콘텐츠 콜렉션(Content Collections)을 위한 프로덕션 검증된 설정을 제공합니다. 콘텐츠 콜렉션은 Markdown/MDX 파일을 Zod 검증이 포함된 타입 안전한 데이터 콜렉션으로 변환해주는 TypeScript 최우선 도구입니다. 블로그, 문서 사이트 또는 콘텐츠 중심의 Vite + React 애플리케이션을 구축할 때 타입 안전성과 자동 콘텐츠 검증을 보장하기 위해 사용하세요. Vite 플러그인 구성과 MDX 컴파일부터 배포 최적화 및 스키마 검증에 이르기까지 모든 것을 다룹니다.
polymarket
메타이 스킬은 개발자들이 Polymarket 예측 시장 플랫폼을 활용한 애플리케이션을 구축할 수 있도록 지원하며, 거래 및 시장 데이터를 위한 API 통합 기능을 포함합니다. 또한 WebSocket을 통한 실시간 데이터 스트리밍을 제공하여 실시간 거래와 시장 활동을 모니터링할 수 있습니다. 이를 통해 거래 전략을 구현하거나 실시간 시장 업데이트를 처리하는 도구를 생성하는 데 활용할 수 있습니다.
creating-opencode-plugins
메타이 스킬은 개발자들이 명령어, 파일, LSP 작업 등 25개 이상의 이벤트 유형에 연결되는 OpenCode 플러그인을 만들 수 있도록 돕습니다. JavaScript/TypeScript 모듈을 위한 플러그인 구조, 이벤트 API 명세, 구현 패턴을 제공합니다. OpenCode AI 어시스턴트의 라이프사이클을 사용자 정의 이벤트 기반 로직으로 가로채거나, 모니터링하거나, 확장해야 할 때 사용하세요.
sglang
메타SGLang은 RadixAttention 프리픽스 캐싱을 활용하여 JSON, 정규식, 에이전트 워크플로우를 위한 고속 구조화 생성에 특화된 고성능 LLM 서빙 프레임워크입니다. 특히 반복되는 프리픽스가 있는 작업에서 상당히 빠른 추론 속도를 제공하여 복잡한 구조화 출력 및 다중 턴 대화에 이상적입니다. 제약 디코딩이 필요하거나 광범위한 프리픽스 공유가 있는 애플리케이션을 구축할 때는 vLLM과 같은 대안보다 SGLang을 선택하십시오.
