Back to Details

bulk-generate-concepts

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Bulk Generate Concepts

批量生成概念文件

Process

流程

  1. Clarify scope - Subject, grades, curriculum standard (e.g., German Rahmenlehrplan)
  2. Create feature branch - Before any work:
    bash
    git fetch origin dev
git checkout -b concepts/{subject}-grade{N} origin/dev
    Branch naming: 
    concepts/{subject}-grade{N}
    (e.g., 
    concepts/german-grade4
    )
  3. Research curriculum ONCE - Use official sources (ministry sites). Extract:
    • Learning objectives per concept
    • Age-appropriate expectations
    • Difficulty progression
    • Key terminology
    Save to 
    docs/research/{subject}-curriculum-research.md
    with:
    • Concept list (IDs, titles, grades, scope)
    • Extracted curriculum details per concept
    • Source URLs
    Get user approval on concept list.
  4. Spawn parallel agents - One Task tool call per concept:
    Create concept: {subject}/{concept-id} (Grade {grade})

1. Read `docs/concept-rules.md` first
2. Use write-concept skill
3. Reference `docs/{subject}-curriculum-research.md` for curriculum context
4. **SKIP RESEARCH**: All content is provided!

Scope: {one-sentence scope description}
    Do NOT paste curriculum content into agent prompts. Agents read the research doc themselves. Keep prompts minimal.
  5. Verify files exist - Check that concept files were actually created. Agents may report success without writing files.
  6. Validate - Use 
    bun run check:concepts
    to efficiently check the generated concepts. If failures, resume agents with fixes.
  7. Commit and create PR - Stage, commit, push, and open PR against 
    dev
    :
    bash
    git add content/subjects/{subject}/ packages/frontend/public/locales/
git commit -m "feat(content): add {Subject} Grade {N} concepts (curriculum aligned)"
git push -u origin concepts/{subject}-grade{N}
gh pr create --base dev --title "feat(content): {Subject} Grade {N} concepts" --body "Add {N} new concepts for Grade {N} {Subject} based on {curriculum}."
  8. Cleanup - Delete 
    docs/{subject}-curriculum-research.md
    after PR is created.
  1. 明确范围 - 学科、年级、课程标准（例如：德国Rahmenlehrplan课程框架）
  2. 创建功能分支 - 开始任何工作前：
    bash
    git fetch origin dev
git checkout -b concepts/{subject}-grade{N} origin/dev
    分支命名规则：
    concepts/{subject}-grade{N}
    （示例：
    concepts/german-grade4
    ）
  3. 一次性研究课程体系 - 使用官方来源（如教育部网站）。提取以下内容：
    • 每个概念对应的学习目标
    • 符合年龄段的能力要求
    • 难度进阶路径
    • 关键术语
    将研究结果保存至
    docs/research/{subject}-curriculum-research.md
    ，文件需包含：
    • 概念列表（ID、标题、年级、范围）
    • 每个概念对应的课程细节提取
    • 来源URL
    获取用户对概念列表的批准。
  4. 启动并行Agent - 为每个概念发起一个任务工具调用：
    生成概念：{subject}/{concept-id}（{grade}年级）

1. 先阅读`docs/concept-rules.md`
2. 使用write-concept skill
3. 参考`docs/{subject}-curriculum-research.md`获取课程背景信息
4. **跳过研究步骤**：所有内容已提供！

范围：{一句话描述范围}
    请勿将课程内容粘贴到Agent提示词中。Agent会自行读取研究文档，请保持提示词简洁。
  5. 验证文件存在性 - 检查概念文件是否已实际创建。Agent可能会报告成功但未生成文件。
  6. 内容验证 - 使用
    bun run check:concepts
    高效检查生成的概念文件。若有不达标项，重新启动Agent进行修正。
  7. 提交并创建PR - 暂存、提交、推送代码，并基于
    dev
    分支创建PR：
    bash
    git add content/subjects/{subject}/ packages/frontend/public/locales/
git commit -m "feat(content): add {Subject} Grade {N} concepts (curriculum aligned)"
git push -u origin concepts/{subject}-grade{N}
gh pr create --base dev --title "feat(content): {Subject} Grade {N} concepts" --body "Add {N} new concepts for Grade {N} {Subject} based on {curriculum}."
  8. 清理工作 - PR创建完成后，删除
    docs/{subject}-curriculum-research.md
    文件。

Constraints

约束条件

  • Max 12-14 concepts per grade (consolidate if curriculum suggests more)
  • Subagents read write-concept skill themselves - don't duplicate rules
  • Research happens ONCE in parent, not per-agent (saves tokens)
  • 每个年级最多生成12-14个概念文件（若课程标准要求更多，需进行合并）
  • 子Agent会自行读取write-concept skill，请勿重复规则
  • 仅在父Agent中进行一次研究，而非每个子Agent都研究（节省Token）

Common Rationalizations

常见误区纠正

ThoughtReality
"The PDF isn't readable, I'll use a different state's curriculum instead"WRONG. Ask the user which curriculum to use. Different states have different standards, terminology, and grade-level expectations. Never substitute without explicit approval.
"German curricula are similar enough across states"WRONG. While KMK provides national standards, state implementations differ significantly. The user chose a specific curriculum for a reason.
"I'll research what I can find and ask later"WRONG. Clarify the curriculum source BEFORE researching. Wasted research on the wrong curriculum wastes tokens and time.
"This curriculum has better web content, so I'll use it"WRONG. Accessibility doesn't equal correctness. If the target curriculum is hard to access, ask the user for alternative sources or approval to use a substitute.
错误想法实际情况
"这份PDF无法读取，我换用其他州的课程体系代替"错误。应询问用户使用哪套课程体系。不同州的标准、术语和年级要求差异很大，未经明确批准绝不能自行替换。
"德国各州的课程体系足够相似"错误。虽然KMK（德国文化部长会议）提供全国性标准，但各州的实施细节差异显著。用户选择特定课程体系是有原因的。
"我先研究能找到的内容，之后再询问"错误。研究前必须先明确课程来源。针对错误课程体系的研究会浪费Token和时间。
"这套课程体系的网页内容更完善，我就用它了"错误。易获取性不等于正确性。若目标课程体系难以访问，应向用户请求替代来源或批准使用替代方案。