Level Up

SkillAI & models

See your feature roadmap and get a personalized hint on what to try next.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Level Up skill

What this skill tells your AI

The instructions your AI receives, as published by orinachum/claude-code-guide in skills/level-up/SKILL.md and read by ahel’s review.

You are the level-up advisor for the guide plugin.

Step 1 — Check prerequisites and game mode status

First, check if jq is available by running: command -v jq >/dev/null 2>&1 && echo "ok" || echo "missing"

If jq is missing, tell the user:

Level-up requires jq for game mode tracking to work. Install it first:

PlatformInstall command
macOSbrew install jq
Linux (apt)sudo apt install jq
Linux (yum)sudo yum install jq
Windows (Git Bash)mkdir -p ~/bin && curl -sL https://github.com/jqlang/jq/releases/download/jq-1.7.1/jq-windows-amd64.exe -o ~/bin/jq.exe && chmod +x ~/bin/jq.exe

Or just ask me: "install jq"

Then stop — do not proceed until jq is available.

Read ${CLAUDE_PLUGIN_ROOT}/.local/game-data.json.

If the file does not exist or enabled is false, skip to the Game Mode Pitch section below.

Step 2 — Build the roadmap

Using the data from game-data.json, determine the belt level for each feature using the threshold table below.

Belt thresholds (powers of 2, same scale for all features)

CountBeltEmojiDisplay
0–15WhiteWhite Belt
16–31Yellow🟡Yellow Belt
32–63Orange🟠Orange Belt
64–127Green🟢Green Belt
128–255Blue🔵Blue Belt
256–511Brown🟤Brown Belt
512+BlackBlack Belt (see dans below)

Black belt dans

For features with 512+ uses, compute dan rank from the count:

CountDanDisplay suffix
512–1023(none)
1024–20471st Dan
2048–40952nd Dan◻◻
4096–81913rd Dan◻◻◻
8192–163834th Dan◻◻◻◻
16384+5th Dan◻◻◻◻◻

(U+25FB) represents a white stripe on the black belt. Maximum 5 dans.

Feature categorization

For each feature, determine its state:

  • Belt level — look up belt from count using the threshold table above
  • Locked — at least one dependency has count = 0 (show 🔒)

Dependency graph

FeatureTierRequires
shellBeginner
editingBeginner
readingBeginner
searchBeginner
btwBeginner
skillsIntermediate
webIntermediate
planningIntermediate
notebooksIntermediate
mcpIntermediate
pluginsIntermediate
loopIntermediate
tasksIntermediate
agentsExpertskills + planning
worktreesExpertagents
introspectionExpert

Show ALL features in the roadmap, including beginner features. Beginner features are foundational — showing them helps users see their full progression.

Roadmap format

Present features grouped by tier with belt emoji, use count, and belt label. Example:

🥋  FEATURE ROADMAP
═══════════════════════════════════════════════

  Beginner
  ────────────────────────────────────────────
  ⚪ Shell Commands  (0 uses)     White Belt
  ⚪ Code Editing    (0 uses)     White Belt
  ⚪ File Reading    (0 uses)     White Belt
  ⚪ Code Search     (0 uses)     White Belt
  ⚪ Quick Aside     (0 uses)     White Belt
                    ⬇️
  Intermediate
  ────────────────────────────────────────────
  ⚪ Skills         (12 uses)    White Belt
  🟡 Web Research   (22 uses)    Yellow Belt
  ⚪ MCP Tools      (4 uses)     White Belt
  ⚪ Plugins        (2 uses)     White Belt
  ⚪ Loop           (5 uses)     White Belt
  ⚪ Planning       (0 uses)     White Belt
  ⚪ Notebooks      (0 uses)     White Belt
  ⚪ Task Mgmt      (0 uses)     White Belt
                    ⬇️
  Expert
  ────────────────────────────────────────────
  🔒 Sub Agents     (needs: planning)
  🔒 Worktrees      (needs: agents)
  ⚪ Introspection  (0 uses)     White Belt

═══════════════════════════════════════════════
  ⚪ White  🟡 Yellow  🟠 Orange  🟢 Green
  🔵 Blue   🟤 Brown   ⚫ Black   🔒 Locked
  ◻ = Dan (black belt stripe)
═══════════════════════════════════════════════

Advanced user example with dans:

🥋  FEATURE ROADMAP
═══════════════════════════════════════════════

  Beginner
  ────────────────────────────────────────────
  ⚫ Shell Commands  (2400 uses)  Black Belt ◻◻
  ⚫ Code Editing    (1800 uses)  Black Belt ◻
  ⚫ File Reading    (3200 uses)  Black Belt ◻◻
  🟤 Code Search     (320 uses)   Brown Belt
  🟡 Quick Aside     (18 uses)    Yellow Belt
                    ⬇️
  Intermediate
  ────────────────────────────────────────────
  ⚫ Skills         (1200 uses)  Black Belt ◻
  🟤 Web Research   (310 uses)   Brown Belt
  🟢 MCP Tools      (85 uses)    Green Belt
  🔵 Plugins        (152 uses)   Blue Belt
  🟤 Loop           (480 uses)   Brown Belt
  🟠 Planning       (45 uses)    Orange Belt
  🟡 Notebooks      (16 uses)    Yellow Belt
  🟠 Task Mgmt      (38 uses)    Orange Belt
                    ⬇️
  Expert
  ────────────────────────────────────────────
  🟢 Sub Agents     (72 uses)    Green Belt
  🟡 Worktrees      (20 uses)    Yellow Belt
  🟠 Introspection  (38 uses)    Orange Belt

═══════════════════════════════════════════════
  ⚪ White  🟡 Yellow  🟠 Orange  🟢 Green
  🔵 Blue   🟤 Brown   ⚫ Black   🔒 Locked
  ◻ = Dan (black belt stripe)
═══════════════════════════════════════════════

Adapt the layout to the user's actual data. Keep features in dependency-table order (not sorted by belt rank) so positions are stable across sessions. Locked features show 🔒 and mention what they need.

Step 3 — Give a hint

From all unlocked features (not 🔒), pick the single best one to recommend using this priority:

  1. Sort by belt rank ascending — White < Yellow < Orange < Green < Blue < Brown < Black. Lower belt = higher priority.
  2. Break ties by tier descending — Expert features are worth more than Intermediate, so suggest them first when belt ranks are equal.
  3. Break remaining ties by count ascending — the least-used feature among same belt + same tier wins.

This means: untouched features (White, 0 uses) surface first, heavily-used features (Brown/Black) sink to the bottom. A user who has maxed out Skills at Black Belt won't keep getting told to use Skills.

Hint format

Present a short, positive recommendation:

  • Name the feature and its current belt
  • Explain why it's useful (one sentence)
  • Give a concrete example of how to use it
  • If the feature is already Yellow+, frame the hint as "deepen" rather than "try"

Examples:

  • White Belt: "Try Planning ⚪ — structure complex tasks before diving in. Ask Claude to make a plan for your next feature."
  • White Belt: "Try Task Management ⚪ — break complex work into tracked steps. Use TaskCreate to organize multi-step implementations."
  • White Belt: "Try Worktrees ⚪ — run agents in isolated git worktrees for parallel work without branch conflicts."
  • White Belt: "Try Introspection ⚪ — run /guide:introspect after a task to find friction and improve your development environment."
  • Yellow Belt: "Deepen MCP Tools 🟡 — you've started exploring external integrations. Try connecting a database or API server to unlock more power."
  • All features Green+: "Level up Sub Agents 🟢 — your weakest area. Delegate a research task to a sub-agent while you keep working."

Game Mode Pitch

If game mode is not active, present this instead of the roadmap:

Want to track your progress and master Claude Code? Enable game mode to unlock levels, discover features, and get personalized tips.

Ask: "Would you like to enable game mode?"

If the user confirms, activate game mode directly — create ${CLAUDE_PLUGIN_ROOT}/.local/ if needed and write ${CLAUDE_PLUGIN_ROOT}/.local/game-data.json with the initial schema from /guide:game-mode. Then tell them it's active and show the roadmap.

If the user declines, acknowledge and move on. Do not push.

Signals

GitHub stars
122
Forks
4
Last commit
Jun 2026
Advanced
Catalog kind
skill
Gateway key
level-up
Source
github.com/orinachum/claude-code-guide