learning-project-planner

name	learning-project-planner
description	Use when a user wants to turn a study topic into a durable self-study project that future agent sessions can resume, with roadmap, context, dashboard, and first-week lesson artifacts.

Learning Project Planner

Use this skill to turn a topic into a durable learning project, not a one-off study plan. The output should preserve long-term context so future agent sessions can resume the learning state quickly.

Default Project Shape

<topic>/
├── <slug>-learning-roadmap.md
├── <slug>-learning-context.md
├── <slug>-learning-dashboard.html
└── week-01-<first-topic>.md

Workflow

1. Clarify only if necessary. If the user gives a topic and broad intent, proceed with reasonable assumptions.
2. Inspect nearby existing learning projects when available, especially user-specific examples under ~/Documents.
3. Build the learning route around the user's purpose, not around a textbook table of contents.
4. Generate four artifacts: roadmap, context, dashboard, and first-week lesson.
5. Keep the first version compact enough to use immediately; leave room for future weekly notes.
6. Verify files exist and basic dashboard syntax is sane.

Roadmap Requirements

• one-paragraph purpose
• 4-6 phases
• each phase has time, goal, themes, output standard
• a week-by-week table
• a fixed weekly template
• suggested prompts the user can say to the agent later

Good week rows use this shape:

| Week | Core question | Tool / anchor material | Exercise output |

Choose the duration based on topic size:

• 6-8 weeks for compact tool or paper tracks
• 12 weeks for focused technical topics
• 18-24 weeks for cross-disciplinary or worldview topics
• 24+ weeks only when the user clearly wants a long course

Context File Requirements

Create <slug>-learning-context.md with:

• update date
• total course objective
• file list
• current progress
• user current understanding state
• user strengths and risks
• teaching strategy
• key distinctions
• relation to nearby learning projects

This file is for future agent sessions. Write it as persistent state, not marketing copy.

Week 01 Requirements

Create week-01-<topic>.md with:

• the first core question
• one intuitive explanation
• one concrete example
• one connection to the user's purpose
• 2-4 exercises
• 3-5 review questions

Do not overload Week 01 with the whole field. Its job is to start the track.

Dashboard Requirements

Create <slug>-learning-dashboard.html as a local single-file dashboard with:

• progress bar
• phase cards
• week checklist
• anchor resources
• notes and questions textareas
• localStorage persistence

Prefer using scripts/create_learning_dashboard.py when generating a dashboard from structured phase/week lists.

Example:

python3 scripts/create_learning_dashboard.py \
  --title "Example Learning Dashboard" \
  --subtitle "A short purpose statement." \
  --slug "example" \
  --phases-json '[{"title":"1. Foundations","body":"Core concepts and first exercises."}]' \
  --weeks-json '[{"question":"Why does this matter?","tool":"Core distinction","output":"Write a five-sentence explanation"}]' \
  --resources-json '[{"title":"Anchor paper","body":"Read after Week 2."}]' \
  --output "/path/to/example-learning-dashboard.html"

Teaching Bias

• Prefer core questions over chapter lists.
• Use papers, textbooks, videos, and docs as anchors, not the whole learning loop.
• Include one small exercise per week.
• Keep technical, philosophical, and product layers separate when the topic spans them.

Common Failure Modes

• Producing only a chat answer instead of files.
• Overloading Week 01 with the whole field.
• Writing a dashboard that has no persistent notes or progress state.
• Losing the user's purpose and copying a generic syllabus.

Existing User Patterns

When useful, inspect these projects as style anchors:

• /Users/tianjiaowu/Documents/LoRA Learning
• /Users/tianjiaowu/Documents/Quantum Physics
• /Users/tianjiaowu/Documents/Enactive AI Learning

Use references/pattern.md for the distilled pattern if you need a quick reminder.