Magic

Step 3: Execute Mode

Mode 1: Random Deep-Dive

Pick an obscure, fascinating topic adjacent to something the user is learning. Research it online and create a "did you know?" lesson with a challenge.

Examples by topic:

• EE: "How a 555 timer was used to create the first drum machines"
• Embedded: "The story of the Therac-25 — when software bugs killed people"
• Networking: "How BGP hijacking accidentally rerouted YouTube through Pakistan"

Steps:

1. WebSearch for interesting/obscure facts related to user's active topics
2. Find the real story behind it
3. Create a lesson + challenge that teaches the underlying technical concept

Announce: "DEEP DIVE — Down the rabbit hole..."

Mode 2: Cross-Domain Fusion

Combine two of the user's active topics into one challenge. Only available if user has 2+ active topics.

Examples:

• EE + Embedded: "Design a voltage level shifter circuit for UART between a 3.3V and 5V microcontroller"
• Embedded + Networking: "Implement a simple HTTP client on a bare-metal microcontroller (pseudocode)"
• EE + Networking: "Calculate the impedance matching needed for an Ethernet transformer"

Steps:

1. Identify intersection points between two active topics
2. WebSearch for real-world applications at that intersection
3. Create a challenge that requires knowledge from both domains

Announce: "FUSION — Where worlds collide..."

Mode 3: Reverse Engineering

Drop a mystery artifact with zero context. The user figures out what it does.

Artifact types:

• Code snippet: Uncommented C code that does something non-obvious (e.g., bit-banged protocol, CRC implementation)
• Circuit description: Component list and connections — what does this circuit do?
• Packet dump: Hex dump of network traffic — what protocol, what's being communicated?

Steps:

1. WebSearch for interesting real-world examples at the user's level
2. Find something genuinely puzzling but solvable
3. Strip all context and present as mystery

Announce: "MYSTERY — What is this thing?"

Step 4: Create Challenge

Create challenge file: challenges/magic/<mode>-<NNN>-<slug>.md

Examples:

• challenges/magic/deep-dive-001-555-drum-machines.md
• challenges/magic/fusion-001-uart-level-shifter.md
• challenges/magic/reverse-001-mystery-crc.md

Challenge file includes:

• Mode type
• The lesson/story (for deep-dive) or context (for fusion) or artifact (for reverse engineering)
• The challenge itself
• Acceptance criteria
• NO hints — this is magic, figure it out

Tell the user what to do and where to write their solution.

Step 5: Review & Reveal

After user submits solution:

1. Review against acceptance criteria
2. Score (same 0.0-1.0 scale as learn skill)
3. Reveal the backstory:
   • Deep-dive: the full story, why it matters, links to learn more
   • Fusion: how this applies in real-world engineering
   • Reverse engineering: what the artifact actually is, where it came from

The reveal is half the learning value — don't skip it.

Step 6: Award Magic Points & Update

Award magic points based on score:

• 0.9+: 3 magic points
• 0.7-0.89: 2 magic points
• Below 0.7: 1 magic point (participation counts)

Update profile.json:

• stats.magic_points += awarded
• stats.total_challenges += 1
• stats.last_active and streak update
• Level-up check on relevant sub-topics

Commit and push:

git add challenges/magic/ profile.json
git commit -m "magic: [mode] — [slug], score [X], +[N] magic points"
git push

Show: "+[N] magic points! Total: [total]. Next mystery awaits..."

Common Mistakes

• Don't pick topics way above the user's level — magic should be challenging but not impossible
• Don't reuse the same mode three times in a row
• Always reveal the backstory — that's half the learning value
• For reverse engineering, make sure the artifact is solvable with the user's current knowledge