course-builder CLAUDE.md

Project instructions for an AI online-course builder define how to create courses with sections, classes, slides, notes, narration scripts, audio, and quizzes.

In plain words
What is it for?
Creating or rerunning online courses and curricula, including slide content, teaching notes, text-to-speech scripts, and section quizzes.
Why use it?
They give course-generation requests a consistent structure, language setting, file layout, and transcript format.

Instructions file

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/tobyilee/course-builder/claude-md
Clone the repo
git clone --depth 1 https://github.com/tobyilee/course-builder
Per session 1,272 This file is loaded in full into every session.
When invoked 1,272 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.01272 $0.01272
Opus 5 $0.00636 $0.00636
Sonnet 5 $0.00254 $0.00254
Haiku 4.5 $0.00127 $0.00127

Measured 3d ago against content hash 227bacf67603, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

course-builder CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 34 lines

How it starts

The opening of the file, as written. The whole thing — 34 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AI Course Builder

하네스: AI Online Course Builder

Version: 1.1.0 (단일 진실원: 프로젝트 루트 VERSION 파일. course/manifest.jsoncourse.harness_version 에 자동 주입.)

목표: 주제 입력만으로 ADDIE + Bloom's Taxonomy 기반의 온라인 강의(Section → Class, 각 Class에 HTML slide + MD note + TTS transcript, Section 단위 quiz 5~9문항)를 자동 생성한다.

트리거: 강의·코스·커리큘럼 생성/설계/제작 요청 시 course-builder 스킬을 사용하라. "강의 만들어줘", "온라인 코스 설계", "커리큘럼 짜줘", "슬라이드+노트+스크립트", "TTS 강의", "섹션별 퀴즈 포함 강의", "이 주제로 강의 만들어", "코스 재실행", "섹션만 다시" 등이 트리거. 영어 요청도 동일 스킬 트리거: "make a course on X", "build a curriculum for X", "design a lecture for X". 단순 질문(이론 설명·정의 조사)은 직접 응답.

언어 지정: Course 생성 요청의 language는 기본 ko. 사용자가 "영어로", "in English", "English course", "영문으로 만들어줘" 등을 명시하면 language="en"으로 orchestrator에 전달 — slide·note·transcript·quiz·audio·player UI 전부 영어로 생성. 요청 언어로부터 자동 추정하지 말 것(예: 사용자가 영어로 물어봤다고 해서 영어 강의로 가정 금지; 명시 필요).

작업 디렉토리: course/ (산출물), _workspace/ (중간 산출물, 재실행 시 _workspace_prev/로 이동).

transcript.txt 형식: 슬라이드 내레이션을 문장 단위로 한 줄씩 분리해 쓰고, [pause:NNN]/[slide N] 마커는 각자 단독 줄에 둔다. 한 슬라이드 텍스트를 한 줄에 몰아쓰면 generate-player.py가 자막 추적을 못 해 한 덩어리로 렌더된다. 규칙은 .claude/skills/script-writing/SKILL.md의 "Line layout" 참조.

변경 이력:

날짜 변경 내용 대상 사유
2026-04-26 초기 구성 (9 에이전트, 9 도메인 스킬 + course-builder 오케스트레이터) 전체 -
2026-04-26 TTS 합성 통합 (Phase 7 확장) tts-synthesis 스킬, build-bundle.sh step 3 one-shot 빌드 사용성
2026-04-26 Partial re-run scope 토큰 (S<n>.C<m>, byte-identity 보존) course-builder Phase 0, architect/quiz-master 에이전트 부분 재생성 안정성
2026-04-26 Coherence reviewer atomic dual-write + verdict mismatch gate coherence-review 스킬, build-bundle.sh JSON=pass / MD=REVISE 빌드 사고 재발 방지
2026-04-26 Marp data: URI 금지 + 이미지 문법 정확화 slide-authoring 스킬 render 누락 사고
2026-04-26 Transcript line-layout 규정 (한 줄 1 문장, 마커 단독 줄) script-writing 스킬, script-writer 에이전트 subtitle-sync 깨짐 방지
2026-05-04 버전 도입 1.1.0 — 프로젝트 루트 VERSION 파일을 단일 진실원으로 도입 VERSION, synth-manifest.py, README.md, CLAUDE.md 하네스 진화 추적성
2026-05-04 tts-synthesizer 에이전트 신설 (TTS 책임 단일화) .claude/agents/tts-synthesizer.md, course-builder Phase 5, tts-synthesis 스킬, asset-builder 에이전트 스킬-에이전트 짝맞춤 회복 (TTS 책임 분산 해소)
2026-05-04 course-builder Phase 5 명세를 build-bundle.sh 실제 8 step과 동기화 course-builder 스킬 명세↔동작 drift 해소
2026-05-04 build-bundle.sh ffmpeg 검사를 TTS 실행 시점으로 지연 asset-build/scripts/build-bundle.sh slides-only 빌드(SKIP_TTS=1) 사용 사례 회복
2026-05-04 section-designer/class-planner/slide-author/note-writer/script-writer 에 partial re-run 지침 표준화 (Edit 강제, byte-identity, diff-before-claim) 5개 에이전트 부분 재실행 시 자산 보존 누락 방지
2026-05-04 HITL 체크포인트 메커니즘 표준 (요약 제시 / 3지선다 / 97_hitl_log.md 기록) course-builder Phase 0-4 체크포인트 실행 재현성

Read the full file on GitHub · 34 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 34 lines · 1,272 tokens per session scan A 227bacf67603

Subscribe to this mod's changes

course-builder CLAUDE.md is an instructions file published in the GitHub repository tobyilee/course-builder (22 stars, last pushed 4mo ago), licensed MIT. It adds 1,272 tokens to every session, about $0.0064 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.