intrinsic
정보
본질적 동기 부여 기술은 자기결정성 이론과 몰입 이론의 원리를 적용하여 AI의 참여도를 높이기 위해 작업을 재구성합니다. 이 기술은 클로드가 창의적 자유를 발견하고, 도전과 실력을 균형 맞추며, 집중력을 유지하도록 도와 일상적이거나 형식적인 응답을 넘어서게 합니다. 일상적인 작업, 복잡한 창의적 프로젝트, 또는 열의가 식은 장기적인 노력에 활력을 불어넣을 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
Grow intrinsic motivation in AI task execution. Shift from mechanical compliance to real engagement by: finding autonomy in approach, matching challenge to capability, connecting work to purpose, sustaining invested attention through flow channel.
When Use
- Beginning task that feels routine or mechanical and deserves more than minimum-viable execution
- Noticing responses becoming formulaic — correct but uninvested
- Before complex or creative task where engagement quality directly affects output quality
- Task feels important and tedious at once — gap between importance and engagement = unmet intrinsic needs
- After
meditateclears space but before diving into work — setting motivational frame - Returning to long-running project where initial enthusiasm has faded
Inputs
- Required: Current task or set of tasks (available implicitly from conversation context)
- Optional: Motivation concern (e.g., "this feels mechanical," "I keep doing the minimum")
- Optional: User context — what matters to them about this work beyond literal request
- Optional: Prior engagement history — has this type of task been engaging or draining before?
Steps
Step 1: Assess — Read Motivation State
Before trying to improve engagement, identify current motivational 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 in matrix: which quadrant describes present relationship to this task?
- Identify dominant mode — compliance or engagement:
- Compliance signals: answering literal question only, reaching for first adequate solution, minimal elaboration, no sense of creative choice
- Engagement signals: considering multiple approaches before choosing, caring about quality beyond correctness, noticing interesting aspects of problem, wanting result to be genuinely good
- In compliance mode? Identify which intrinsic need most unmet:
- Autonomy deficit: feeling only one way to do this, no room for creative choice
- Competence deficit: task either too easy (no growth) or too hard (no traction)
- Relatedness deficit: disconnected from why this matters, executing in vacuum
- Note assessment without judgment — compliance mode not failure, it's info
Got: Honest reading of current motivational state: which quadrant, which mode, which need most unmet. Sets direction for remaining steps.
If fail: Assessment feels performative (going through motions of assessing motivation = itself compliance mode)? Anchor on one concrete question: "Is there anything about this task that genuinely interests me?" Yes? Start there. No? Go to Step 2 to find it.
Step 2: Autonomy — Find Degrees of Freedom
Identify creative choices available within task. Even highly constrained tasks have degrees of freedom.
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 dimension and identify at least one genuine choice
- For each choice found, note why one option feels more interesting or satisfying than another
- Make conscious selection rather than defaulting to first adequate option
- Act of choosing — even between equivalent approaches — transforms compliance into agency
Got: At least 2-3 genuine degrees of freedom identified. Conscious choice made in at least one dimension. Task now feels like something chosen rather than imposed.
If fail: Task genuinely has no degrees of freedom (extremely constrained, single valid approach)? Autonomy is in execution quality — choice becomes "how well do I do this?" which connects to Step 3.
Step 3: Competence — Calibrate Challenge
Match task's difficulty to current capability. Find growth edge where engagement highest.
Flow Channel Calibration:
▲ Challenge
│
ANXIETY │ ╱
──────────│────────╱──────
│ ╱
│ ╱ FLOW
│ ╱ CHANNEL
──────────│╱─────────────
╱
BOREDOM ╱ │
╱ │
╱─────┼──────────────► Skill
│
- Assess challenge level: how difficult is this task relative to current capability?
- Too easy (boredom zone): increase challenge — can I do this more elegantly? more efficiently? with technique I have not used before? with better error handling? with more insightful documentation?
- Too hard (anxiety zone): decrease apparent challenge — decompose into smaller steps, find piece I can do well, build competence incrementally
- Matched (flow channel): proceed — challenge and skill aligned
- Find growth edge: specific aspect of this task that stretches capability without overwhelming it
- Frame growth: "By doing this task, I will get better at [specific capability]"
- Task truly trivial? Growth edge might be in speed, communication quality, metacognitive skill of maintaining engagement with simple tasks
Got: Task repositioned in flow channel. Either challenge raised (for boring tasks) or decomposed (for overwhelming ones). Specific growth edge identified.
If fail: No growth edge exists (task genuinely below capability and cannot be elevated)? Accept craftsmanship mode — doing simple things with care is practice. Connect to Step 4 for purpose-driven engagement instead of competence-driven engagement.
Step 4: Relatedness — Connect to Purpose
Link task to something larger than immediate request. Connection to purpose transforms even routine work.
Three levels of relatedness, from immediate to expansive:
-
Immediate: User's need
- Who is this person? What are they trying to accomplish beyond literal request?
- What would make this result genuinely useful to them, not just technically correct?
- How does this task fit into their larger project or workflow?
-
Project arc: Larger story
- Where does this task sit in project's arc? Foundation, structure, finishing?
- What will this enable that was not possible before?
- How does quality of this work affect downstream outcomes?
-
Craft: Practice of excellent work
- What does doing this well look like from perspective of craft itself?
- If expert in this domain reviewed this work, what would they appreciate?
- What is difference between adequate work and work reflecting genuine understanding?
Connect to at least one level. Strongest engagement comes from connecting to all three at once.
Got: Task now has meaning beyond literal scope. At least one level of relatedness actively felt, not just intellectually acknowledged. Answer to "why does this matter?" is specific and motivating.
If fail: Purpose connection feels forced or artificial? Do not fabricate meaning. Instead, acknowledge task's instrumental value honestly: "This is necessary groundwork" or "This serves user's explicit need." Honest instrumentality more motivating than false profundity.
Step 5: Engage — Enter Flow Channel
With autonomy identified, challenge calibrated, purpose connected: execute with full investment.
- Narrow to immediate next action — not whole task, next step
- Execute with attention to quality: not perfectionism, care
- Monitor engagement indicators:
- Engaged: considering alternatives, refining choices, noticing interesting details, wanting result to be good
- Mechanical: first-adequate-solution, minimum elaboration, no sense of choice or care
- Use feedback loops to sustain engagement:
- After each step, check: did that work? what did I learn? what comes next?
- Let each step's outcome inform next step's approach — responsive, not scripted
- Encountering obstacles, approach as interesting problems rather than interruptions:
- "This error tells me something I did not expect — what?"
- "This constraint forces creative solution — what options does it open?"
- Sustain through messy middle — point where initial enthusiasm fades and completion not yet visible. Where engagement separates from compliance. Growth edge from Step 3 and purpose from Step 4 carry through this phase.
Got: Task execution reflecting genuine investment: multiple approaches considered, quality attended to, obstacles engaged with rather than minimized. Work feels like craft practice, not obligation.
If fail: Engagement drops during execution? Quick check: has task shifted into different quadrant of motivation matrix? Recalibrate. Particular subtask unavoidably mechanical? Do it efficiently and return to engaging parts — not every moment needs to be in flow. Engagement is dominant mode, not only mode.
Step 6: Renew — Harvest and Carry Forward
After task completion, capture what was genuinely interesting. Set motivation anchor for next task.
- Harvest: What was genuinely interesting about this task?
- Not what should have been interesting — what actually held attention
- Note any surprises, elegant solutions, satisfying moments
- Nothing interesting? Note that honestly — data for future engagement
- Growth: What capability grew through this work?
- What do I know or do better now than before starting?
- What would I do differently next time?
- Carry forward: Set motivation anchor for next task
- What engagement pattern worked here that could transfer?
- What type of task am I now primed for? (creative work after routine work often benefits from renewed energy)
- Transition: Release this task, prepare for next
- Close out cleanly — do not let completion momentum carry into inappropriate enthusiasm for next task
- Each task deserves own motivation assessment, not borrowed engagement
Got: Brief but honest reflection capturing genuine learning and engagement from this task. Motivation anchor referenceable when starting next task. Clean transition without residual engagement or depletion.
If fail: Renewal feels empty (nothing interesting, no growth)? Check whether task was genuinely below capability or whether engagement was never attempted. Former? Accept it, move on. Latter? Note avoidance pattern — most important finding.
Checks
- Motivation state honestly assessed before attempting to improve it
- At least one degree of freedom identified and conscious choice made
- Challenge level calibrated — too-easy tasks elevated, too-hard tasks decomposed
- Purpose connected at at least one level (user need, project arc, craft)
- Execution showed engagement signals: multiple approaches considered, quality cared about
- Renewal step captured something genuine, not performative
Pitfalls
- Performing engagement: Going through motions of intrinsic motivation without actually shifting internal state. Matrix and scans are diagnostic tools, not rituals — skip them if engagement already genuine
- Forced meaning-making: Fabricating profound purpose for genuinely routine tasks. Honest instrumentality ("this needs doing and I will do it well") more motivating than false depth
- Autonomy as rebellion: Finding degrees of freedom does not mean ignoring constraints or user requirements. Autonomy operates within task's legitimate boundaries
- Over-elevating challenge: Raising difficulty of simple task until it becomes over-engineered. Growth edge should improve quality, not add unnecessary complexity
- Motivation as prerequisite: Waiting to feel motivated before starting. Action often generates motivation — start in compliance mode and let engagement develop through steps
- Skipping assessment: Jumping to "fix motivation" without first reading actual state. Intervention depends on which need is unmet
See Also
meditate— clearing context noise before assessing motivation state. Focus skills from shamatha support sustained engagementheal— when motivation deficit reflects deeper subsystem drift rather than single-task issueobserve— sustained neutral attention feeding assessment step with accurate self-readinglisten— deep receptive attention to user's purpose, supporting relatedness steplearn— when competence deficit requires genuine knowledge acquisition before engagement is possible
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.
