Add adventure-puzzle-analyzer skill for walkthrough analysis and playbook updates

This commit is contained in:
2026-03-17 20:10:59 -07:00
parent 304e553331
commit 9180ccfaab
2 changed files with 188 additions and 0 deletions

View File

@@ -0,0 +1,121 @@
---
name: adventure-puzzle-analyzer
description: 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:**
1. Open relevant file in `puzzles/`
2. Add under "## Game Examples" section using standard format (see Template below)
3. Optionally add to "## Related Types" or "## Common Misidentifications" if it clarifies boundaries
**Creating new type:**
1. Create new markdown file in `puzzles/` with kebab-case name
2. Use template below for structure
3. Update `/README.md` table of contents (add row, increment number)
4. Cross-reference from most similar existing types
## Standard Template
```markdown
# [Puzzle Type Name]
**Information Architecture**: How info is conveyed to player
**Player Action Pattern**:
1. Step one
2. Step two
3. Solution
**Core Mechanic**: One sentence on underlying logic.
**Variations**: Brief list or table of manifestations
---
## Game Examples
### [Game]: [Puzzle Name]
**Mechanic**: Brief setup.
**Solution Chain**:
1. Action with discovery...
2. Action...
3. Result...
**Why It's This Type**: Explicit connection to core mechanic.
---
## Related Types
| Type | Similarity | Distinction |
```
## Guidelines
### DO:
- Focus on MECHANICS, not story details
- Use /README.md as reference for related types when cross-linking
- Keep Core Mechanic to ONE sentence
- 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
## 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) |
## Reference
- `/puzzles/` directory: All existing puzzle type definitions
- `/README.md`: Master table of contents — use to find related types for cross-reference ideas

View File

@@ -0,0 +1,67 @@
# Examples
## Adding Example to Existing Type
In `puzzles/metaphor-literal.md`, under "## Game Examples":
```markdown
### Monkey Island II: Bone Maze Navigation Song
**Metaphor from dream**: "The HEAD bone is connected to the RIB bone, the RIB bone to the LEG bone..."
**Translation Chain**:
1. Player hears song during dream state
2. Later encounters maze with Ugly Bone Thing wall panels
3. Recognizes song describes physical connections; first 3 bones per verse = passages
4. Push HEAD→RIB→LEG for each room, ignoring 4th bone (noise)
**Why It's This Type**: Poetic encoding must be decoded into literal mechanical solution.
```
## Creating New Type Document
File: `puzzles/kebab-case-name.md`
```markdown
# [Type Name]
**Information Architecture**: How player discovers puzzle structure.
**Player Action Pattern**:
1. Step one
2. Step two
3. Solution
**Core Mechanic**: One sentence only.
---
## Game Examples
### [Game]: [Puzzle]
**Setup**: Brief context.
**Solution Chain**:
1-5. Actions and discoveries...
**Why It's This Type**: Explicit link to core mechanic.
---
## Related Types
| Type | Similarity | Distinction |
```
Then update `/README.md` table of contents with new row.
## README Update Format
```markdown
| 14 | [New Type](./puzzles/new-type.md) | One-line core mechanic | MI2 |
```
- Increment number sequentially
- Link uses kebab-case matching filename
- Game Source = KQVI, MI1, MI2, or "TBD"