SuperAPI AI Directory

day4-wrap-and-analyze

General & other

View repository

Updated 10-15 16:43

About this skill

Day 4: Wrap & Analyze

When this skill is invoked, the following STOP PROTOCOLmust be followed.


Terminology

Key terms used in this skill:

Term Description
session-wrap A skill that organizes and documents work when a coding session ends. "Tidying up your desk before leaving work"
multi-agent A pattern where multiple agents work simultaneously. "Receiving reports from each team lead at the same time in a meeting"
Parallel Processing multiple tasks at the same time. "Receiving reports from 4 team leads all at once" (opposite: sequential = one at a time in order)
2-Phase Pipeline First analyze (Phase 1, parallel) → then verify (Phase 2, sequential). "Gather expert opinions, then the team lead double-checks"
frontmatter At the very top of the skill file, ---A "name tag" written wrapped in. The skill's name and description are written here
history-insight A skill that analyzes past session records and extracts insights
session-analyzer An analysis tool that verifies whether a skill ran as intended
Plugin A collection of skills installed from outside. A "skill package" with multiple skills bundled together
fetch Fetching data from outside. "Like a delivery order—just give the URL and the content arrives"
digest Digesting (summarizing, quizzing, learning) the fetched content. "Not just reading and finishing, but chewing it to make it your own"
Skill Chaining Connecting one skill's output as another skill's input. "fetch → digest pipeline"
Quiz-First A learning method where you solve quizzes first without looking at the summary. 9-12% memory improvement effect
compound A skill that records verified insights from work as structured documents. "Organize what you learned in notes and search later"
team-assemble A skill that splits complex work into expert teams and runs them in parallel. "Automatically forming a TF team per project"
content-digest A skill that digests fetched content using the Quiz-First method. "Don't just read—internalize through quizzes"

STOP PROTOCOL — Never Violate

This protocol is the top-priority rule of this skill. Violating the rules below will ruin the class.

Each block must proceed over exactly 2 turns

┌─ Phase A (첫 번째 턴) ──────────────────────────────┐
│ 1. references/에서 해당 블록 파일의 EXPLAIN 섹션을 읽는다    │
│ 2. 기능을 설명한다                                        │
│ 3. references/에서 해당 블록 파일의 EXECUTE 섹션을 읽는다    │
│ 4. "지금 직접 실행해보세요"라고 안내한다                     │
│ 5. ⛔ 여기서 반드시 STOP. 턴을 종료한다.                    │
│                                                          │
│ ❌ 절대 하지 않는 것: 퀴즈 출제, QUIZ 섹션 읽기             │
│ ❌ 절대 하지 않는 것: AskUserQuestion 호출                  │
│ ❌ 절대 하지 않는 것: "실행해봤나요?" 질문                   │
└──────────────────────────────────────────────────────────┘

  ⬇️ 사용자가 돌아와서 "했어", "완료", "다음" 등을 입력한다

┌─ Phase B (두 번째 턴) ──────────────────────────────┐
│ 1. references/에서 해당 블록 파일의 QUIZ 섹션을 읽는다       │
│ 2. AskUserQuestion으로 퀴즈를 출제한다                     │
│ 3. 정답/오답 피드백을 준다                                 │
│ 4. 다음 블록으로 이동할지 AskUserQuestion으로 묻는다         │
│ 5. ⛔ 다음 블록을 시작하면 다시 Phase A부터.                │
└──────────────────────────────────────────────────────────┘

Key Prohibitions (Never Violate)

  1. Do not call AskUserQuestion in Phase A — After explanation + execution guidance, Stop immediately
  2. Do not give quizzes in Phase A — Read the QUIZ section only in Phase B
  3. Do not ask "Did you try running it?" in Phase A — Wait until the user speaks first
  4. Do not do EXPLAIN + QUIZ at the same time in one turn — Always split into 2 turns

Output official document URL (never omit)

At the start of Phase A for every block, at the top of the corresponding reference file > 공식 문서: URL을 must be output exactly as is.

📖 공식 문서: [URL]
  • If the reference file contains multiple URLs, output all of them
  • Do not summarize or omit URLs

Required phrase at the end of Phase A

At the end of Phase A, you must output the phrase in the form below and Stop:

---
👆 위 내용을 직접 실행해보세요.
실행이 끝나면 "완료" 또는 "다음"이라고 입력해주세요.

After this phrase, do not output any tool calls (including AskUserQuestion) or additional text.


Time estimate guide

Block Topic Estimated time
0 Concept understanding ~10분
1 Creating a skill ~30분
2 Execution & verification ~15분
3 History Insight ~10분
4 Session Analyzer ~15분
5 Content digestion experience + bonus skill + wrap-up ~25분
Total ~105분

It may take 90–110 minutes depending on participant pace. Block 1 is the core block that takes the longest. Block 5 (content digestion) is experience-focused, so it proceeds quickly.


Core strategy: Learn by deconstructing the original skill

Proceed in the following way:

  1. In Block 0, understand the structure of the session-wrap skill and the multi-agent principles
  2. In Block 1, participants directly write the SKILL.md of the session-wrap skill (step-by-step guidance)
  3. In Block 2, run the skill you created yourself and check the results
  4. In Block 3, analyze past session records with history-insight
  5. In Block 4, verify skill execution with session-analyzer
  6. In Block 5, experience the content digestion pipeline (fetch-tweet + content-digest)
  7. In Block 5, experience the content digestion pipeline (fetch-tweet → content-digest) with content-digest
  8. In the Block 5 optional exercise, record insights with compound and experience building an agent team with team-assemble

The original session-wrap is installed in the plugin. Participants create their own version while referring to it.


Block-specific rules

  • Block 0 (Concept understanding): In Phase A, explain the multi-agent concept + guide analysis of the original session-wrap structure → Stop. Quiz in Phase B.
  • Block 1 (Creating a skill): In Phase A, guide how to write SKILL.md step by step → participants write it themselves → Stop. Quiz on the structure of the created skill in Phase B. (Longest block — after completion, encourage with "You've followed along well so far!")
  • Block 2 (Execution & verification): In Phase A, guidance on running the created skill + checking results → Stop. Quiz on execution results in Phase B.
  • Block 3 (History Insight): In Phase A, introduce the history-insight skill + guidance on running it → Stop. Quiz in Phase B.
  • Block 4 (Session Analyzer): In Phase A, introduce session-analyzer + guidance on running it → Stop. Quiz in Phase B.
  • Block 5 (Content digestion experience): In Phase A, explain the concepts of fetch-tweet and content-digest + guide hands-on experience with real tweets + introduce compound/team-assemble → Stop. In Phase B, comprehensive quiz + wrap up the 4-day camp.

References file map

Block File Topic
Block 0 references/block0-concept.md Multi-agent patterns + session-wrap concept
Block 1 references/block1-build-session-wrap.md Build the session-wrap skill yourself
Block 2 references/block2-run-session-wrap.md Run + verify the skill you built
Block 3 references/block3-history-insight.md history-insight hands-on practice
Block 4 references/block4-session-analyzer.md session-analyzer hands-on practice
Block 5 references/block5-content-experience.md Content digestion experience (fetch-tweet + content-digest + bonus skills)

File paths are relative to this SKILL.md. Each reference file consists of the ## EXPLAIN, ## EXECUTE, ## QUIZ section.


Progression rules

  • Proceed one block at a time
  • Move with "next", "skip", or a block number/name
  • Run the skill file created in Block 1 during Block 2
  • In the participant's project, .claude/skills/my-session-wrap/SKILL.mdis created
  • When a question about Claude Code comes up, answer using the claude-code-guide agent (built-in tool). After answering, guide the user step by step so they can follow along themselves, and use AskUserQuestion when asking questions. If you judge the built-in agent's answer to be inaccurate, save the official documentation to a file as curland then read it carefully with the Read tool and answer again with accurate information (do not use WebFetch, as it risks summarization/loss)

Start

When starting the skill, first install the latest curriculumand then select a block.

Step 1: Update to the latest skills (required!)

Many new skills have been added starting from Day 4! Be sure to update with the command below. If you do not update, the skills to be used in today's hands-on practice will not be available.

Output the command below and execute it with Bash:

npx skills add ai-native-camp/camp-2 --agent claude-code --yes

Briefly explain the execution result and show the list of newly added skills:

Newly added skills Description
session-wrap A multi-agent skill that automatically organizes work when a session ends
history-insight Analyzes past session records to extract insights
session-analyzer Verifies whether the skill ran as intended
fetch-tweet Fetches X/Twitter tweets and translates/summarizes them
content-digest Digest and learn content using the Quiz-First approach
compound Accumulate insights discovered during work into structured documents
team-assemble Break down and execute complex tasks with a team of expert agents

Step 2: Select a block

Show the table below and ask with AskUserQuestion where to start.

Block Topic Content
0 Understanding concepts Multi-agent patterns, what is session-wrap?
1 Building skills Write session-wrap SKILL.md yourself
2 Run & verify Run the skill you built + check results
3 History Insight Session History Analysis Practice
4 Session Analyzer Session Execution Verification Practice
5 Content Digestion Experience fetch-tweet + Quiz-First Experience + Wrap-up
AskUserQuestion({
  "questions": [{
    "question": "Day 4: Wrap & Analyze\n\n어디서부터 시작할까요?",
    "header": "시작 블록",
    "options": [
      {"label": "처음부터 (Block 0)", "description": "Multi-agent 패턴과 session-wrap 개념부터 차근차근"},
      {"label": "스킬 만들기 (Block 1)", "description": "개념을 이미 알면 바로 스킬 작성하기"},
      {"label": "실행 & 검증 (Block 2)", "description": "스킬을 이미 만들었으면 실행 + 결과 확인"},
      {"label": "분석 도구 (Block 3~4)", "description": "history-insight와 session-analyzer 실습부터"},
      {"label": "콘텐츠 소화 (Block 5)", "description": "fetch-tweet + content-digest + compound + team-assemble 체험"}
    ],
    "multiSelect": false
  }]
})

After selecting a start block → proceed from Phase A of that block.