- Convert inline bold headers to proper ## headers for Core Mechanic and When to Use - Add Index table to match template - Add Symbol Code Translation and other types to skill best practices
7.6 KiB
7.6 KiB
name, description
| name | description |
|---|---|
| adventure-puzzle-analyzer | Analyzes point-and-click adventure game walkthroughs to identify, categorize, and document puzzle mechanics. Use when analyzing new game walkthroughs or updating the puzzle design handbook. |
Adventure Puzzle Analyzer Skill
Quick Start
1. Read walkthrough, list all puzzle solutions
2. For each: what info + how conveyed + what action solved it?
3. Match to existing types in /puzzles/ OR create new type
4. Document and update README.md table of contents
Instructions
Step 1: Extract Puzzle Solutions
For each walkthrough section, note:
- What player needed to discover
- How information was conveyed (dialogue, environment, text, trial-and-error)
- Action(s) that solved it
- Whether info came from one or multiple sources
Step 2: Classify Against Existing Types
Read /README.md for full list of existing types. Quick reference:
| Type | Key Indicator |
|---|---|
| Pattern Learning / Knowledge Transfer | Learn system in Domain A, apply same rules in Domain B |
| Multi-Faceted Plan | Multiple INDEPENDENT requirements gathered from different sources |
| Meta-Puzzle Construction | Sequential chain where STEP N'S OUTPUT = Step N+1's INPUT |
| Memo Chain | Scattered written fragments that synthesize into solution |
| Metaphor-to-Literal | Idiomatic/poetic language decoded into literal game objects |
| Information Brokerage | NPC exchange network: who wants what from whom |
| Sensory Exploitation | NPC has perceptual weakness/sense; exploit directly |
| Distraction Physics | Manipulate ENVIRONMENT to break NPC blocking pattern |
| Observation Replay | Watch sequence once under blocked conditions, reproduce exactly later |
| Timed Consequence | Story urgency without mechanical timer/deadline |
If none fit: Read existing type definitions in /puzzles/ to confirm no match. Consider new type if at least 2 distinct puzzles share same information architecture.
Step 3: Document
Adding example to existing type:
- Open relevant file in
puzzles/ - Add under "## Game Examples" section using standard format (see Template below)
- Optionally add to "## Related Types" or "## Common Misidentifications" if it clarifies boundaries
Creating new type:
- Create new markdown file in
puzzles/with kebab-case name - Use template below for structure
- Update
/README.mdtable of contents (add row, increment number) - Cross-reference from most similar existing types
Standard Template
Every puzzle type page should follow this structure (target: ≤400 lines):
# [Puzzle Type Name]
## Core Mechanic
(3 sentences max explaining what this pattern achieves)
## When to Use
(When designer should consider this puzzle type)
## Solution Chain
1. (Specific player action)
2. (Specific player action)
3. (Specific player action)
## Examples
### [Game Code]: [Puzzle Name]
**Problem**: One sentence on the obstacle.
**Why It's This Type**: Explicit connection to Core Mechanic.
**Solution**:
1. Step...
2. Step...
---
### [Game Code]: [Puzzle Name]
(Same format)
---
### [Game Code]: [Puzzle Name]
(Same format)
---
## Related Types
| Type | Similarity | Distinction |
|------|------------|-------------|
| [Type A](path.md) | Shared characteristic | Key difference |
| [Type B](path.md) | Shared characteristic | Key difference |
## Index
| Game | Puzzle | Section |
|------|--------|---------|
| MI1 | Swordfight Insults | Examples |
| KQVI | [Puzzle Name] | Inspiration |
Key Rules:
- Core Mechanic: ≤ 3 sentences
- Exactly 3 examples (not 5, not 10)
- Solution Chain: numbered, specific actions (not "solve puzzle")
- "Why It's This Type": explicit connection to Core Mechanic
- Move extended analysis to game inspiration pages, not type definitions
- Use source codes (MI1, KQVI, QFG3) not full game names
Guidelines
DO:
- Focus on MECHANICS, not story details
- Keep Core Mechanic to ≤ 3 sentences
- Use exactly 3 examples with consistent format
- Link extended analysis to game inspiration pages
- Use source codes (MI1, KQVI, QFG3) not full game names
- Distinguish parallel (MFP) vs sequential (Meta-Construction) requirements
DON'T:
- Document game-specific mechanics that don't generalize
- Create new types without 2+ examples OR one very complex example
- Use vague terms ("clever," "creative") — be mechanical
- Exceed 400 lines per page — move details to inspiration pages
Quick Pitfalls
| Confusion | Test |
|---|---|
| Pattern Learning vs Observation Replay | Learn SYSTEM (PL) vs memorize SEQUENCE (OR) |
| MFP vs Meta-Construction | Gather in any order? Yes = MFP, No = Meta-Construction |
| Brokerage vs Sensory Exploitation | Trade ITEM for ITEM (IB) vs exploit PERCEPTION weakness (SE) |
| Page length | If >400 lines, move examples to inspiration pages |
How to Design Great Puzzles of This Type
| Type | Best Practices |
|---|---|
| Pattern Learning | Teach the complete system in the tutorial domain before testing it; ensure identical mechanics between learning and application phases |
| Multi-Faceted Plan | Let players discover requirements through exploration, not explicit checklists; ensure the "aha" moment comes from synthesis |
| Meta-Construction | Make each step's output clearly feed into the next; avoid invisible dependencies |
| Memo Chain | Place fragments in logical locations that reward exploration; ensure assembly is obvious once all pieces found |
| Metaphor-to-Literal | Make the idiomatic expression memorable and culturally clear; ensure the literal mapping is logical within game world |
| Information Brokerage | Give each NPC a distinct personality that explains why they want what they want |
| Sensory Exploitation | State the perceived sense explicitly (character says "I trust only what I hear"); ensure the item genuinely matches that sense |
| Distraction Physics | Make the environmental chain logical and observable before player tries it; ensure cause-effect is visible |
| Observation Replay | Make the observed sequence memorable but not trivially reproducible; add stakes so failure has consequence |
| Timed Consequence | Create urgency through narrative stakes, not arbitrary deadlines; ensure player can always see the countdown |
| Symbol Code Translation | Provide clear visual feedback on EXAMINATION—symbols must be readable; ensure shapes/colors on artifacts match button interfaces exactly; first instance should be a sandbox with no penalty until pattern discovered |
| Truth Revelation | Make the hidden information logically deducible from the reveal action; ensure the item or action used has a natural connection to what's being revealed |
| Cross-Temporal Causality | Make cause-effect between time periods obvious once observed; ensure player can always trace back what they changed and why it worked |
| Multi-Character Coordination | Give each character a distinct role that's immediately clear; ensure simultaneous actions are achievable without tight real-time requirements |
Universal Principles:
- Information conveyance should match the puzzle's difficulty — complex puzzles need multiple hints, simple ones need fewer
- Failure should teach, not punish — when player fails, they should learn something useful
- The solution should feel inevitable once discovered — avoid "moon logic" where solution requires arbitrary knowledge
- Reward the player's investment — if puzzle requires effort, the payoff should feel earned
Reference
/puzzles/directory: All existing puzzle type definitions/README.md: Master table of contents — use to find related types for cross-reference ideas