Game Mod Balance Scaling

SKILL.md — game-mod-balance-scaling Skill Registry

Skill Registration

Identity

• Name: game-mod-balance-scaling
• Version: 1.0.0
• Type: Harness (orchestrates multiple sub-skills)
• Domain: Modding Balance Engineering & Scaling
• Phase: Production Ready

Description

Production-grade harness for analyzing game mod balance and scaling design. Provides evidence-backed analysis with academic citations, risk disclosure, and actionable recommendations. Orchestrates 5 specialized sub-skills through a structured workflow with quality gates and graceful degradation.

Triggering Phrases

Use this skill when the user asks about:

• Mod balance analysis or evaluation
• Scaling design for game mods
• Power curve analysis or design
• Economy consistency in mods
• Difficulty curves and progression
• Mod compatibility and interactions
• Balance testing and telemetry
• Game mod balance decisions

Architecture

Flexible Agent & Skill Pattern

This skill uses a modular skill-registry pattern with:

• Chain-of-thought routers: Main harness routes inputs through specialized analysis steps
• Specialized sub-agents: Each sub-skill handles a specific domain task
• Quality gate checkpoints: Structured validation at each stage
• Graceful degradation: Fallback chains for missing data

Harness Flow

USER INPUT
    │
    ▼
[Pre-Flight: Language Detection]
    │
    ├─► Step 1: sub-gather-requirements ──► Structured requirements
    ├─► Step 2: sub-evidence-collector ──► Evidence bundle
    ├─► Step 3: sub-core-analysis ──► Balance analysis
    ├─► Step 4: sub-knowledge-updater ──► Academic evidence
    ├─► Step 5: sub-advisor ──► Risk-disclosed conclusion
    │
    ▼
[Quality Gates: U1-U6, G1-G4]
    │
    ▼
DELIVER OUTPUT

Sub-Skill Registry

| Sub-Skill | Purpose | Trigger | Quality Gate | |-----------|---------|---------|--------------| | sub-gather-requirements | Clarify analysis scope | Always first | Object confirmed | | sub-evidence-collector | Fetch authoritative data | After requirements | ≥1 current + 1 authoritative source | | sub-core-analysis | Balance design analysis | After evidence | Functions explicit, balanced | | sub-knowledge-updater | Query academic evidence | After analysis | ≥1 academic source | | sub-advisor | Synthesize recommendation | After knowledge | Valid verdict + disclosure |

Tool Definitions

Tool Schema

{
  "tool_name": "WebSearch",
  "purpose": "Fetch live domain news and reports",
  "parameters": {
    "query": {
      "type": "string",
      "required": true,
      "description": "Search query for domain information"
    },
    "recency_filter": {
      "type": "string",
      "enum": ["oneDay", "oneWeek", "oneMonth", "noLimit"],
      "default": "oneMonth"
    }
  },
  "output": {
    "results": "Array of search results with title, URL, summary",
    "timestamp": "ISO format timestamp"
  }
}

Tool Execution Handlers

All tools are invoked with:

• Timeout protection: 30-second default timeout
• Retry logic: 3 attempts with exponential backoff
• Fallback chains: Primary → Secondary → Knowledge base
• Error handling: Graceful degradation on failure

Input/Output JSON Schemas

Input Schema

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "user_message": {
      "type": "string",
      "description": "User's analysis request"
    },
    "language": {
      "type": "string",
      "enum": ["en", "vi", "auto"],
      "default": "auto"
    },
    "analysis_type": {
      "type": "string",
      "enum": ["combined", "economy", "power_curve", "difficulty"],
      "default": "combined"
    },
    "mod_data": {
      "type": "object",
      "description": "Optional mod configuration data",
      "properties": {
        "name": {"type": "string"},
        "version": {"type": "string"},
        "target_game": {"type": "string"},
        "scope": {
          "type": "object",
          "properties": {
            "items": {"type": "boolean"},
            "stats": {"type": "boolean"},
            "economy": {"type": "boolean"},
            "difficulty": {"type": "boolean"}
          }
        }
      }
    }
  },
  "required": ["user_message"]
}

Output Schema

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "properties": {
    "report": {
      "type": "object",
      "properties": {
        "title": {"type": "string"},
        "date": {"type": "string", "format": "date"},
        "language": {"type": "string"},
        "executive_summary": {"type": "string"},
        "inputs_and_scope": {"type": "object"},
        "evidence_collected": {"type": "array"},
        "analysis": {"type": "object"},
        "scenarios": {"type": "object"},
        "academic_evidence": {"type": "array"},
        "disclosure": {"type": "string"},
        "verdict": {
          "type": "string",
          "enum": ["Balanced Mod", "Conditional", "Unbalanced", "Inconclusive"]
        },
        "key_risks": {"type": "array"},
        "evidence_chain": {"type": "array"},
        "remediation": {"type": "array"}
      },
      "required": ["title", "date", "executive_summary", "verdict", "disclosure"]
    },
    "metadata": {
      "type": "object",
      "properties": {
        "session_id": {"type": "string"},
        "gates_passed": {"type": "array"},
        "gates_failed": {"type": "array"},
        "degradation_level": {"type": "integer", "minimum": 0, "maximum": 4},
        "confidence": {"type": "string", "enum": ["HIGH", "MEDIUM", "LOW"]}
      }
    }
  },
  "required": ["report", "metadata"]
}

Validation & Execution

Skill Resolution

When invoked, the skill:

1. Loads configuration from /config/settings.json
2. Initializes hooks from /config/hooks.json
3. Detects language from input characters
4. Validates inputs against schema
5. Routes through harness steps sequentially

Execution Protocol

Each step:

1. Pre-hooks: Validate inputs, check degradation
2. Execute sub-skill: Invoke via Skill tool
3. Post-hooks: Update state, log metrics
4. Quality gate: Verify output meets requirements
5. Auto-fix: Attempt correction on failure (2 retries max)
6. Escalate: Degrade on persistent failure

Error Handling

| Error Type | Detection | Recovery | Limit | |------------|-----------|----------|-------| | Source timeout | No response 30s | Retry alternate source | 3 | | Invalid input | Schema mismatch | Ask user to confirm | 2 | | Missing input | Field absent | Proceed with available + flag | N/A | | Knowledge miss | No matches | WebSearch gap-fill | 2 | | Gate failure | Validation fails | Auto-fix → degrade | 2 retries |

Quality Gates

Universal Gates (U1-U6)

All outputs must pass:

• U1: ≥3 sources cited, ≥1 academic/authoritative
• U2: Disclosure/limitations before recommendation
• U3: Evidence hierarchy stated per source
• U4: Language matches user preference
• U5: Output uses declared template
• U6: Every claim traceable to source or flagged

Domain Gates (G1-G4)

Mod-specific requirements:

• G1: Scaling functions explicit (linear/poly/log)
• G2: Power/economy curve balanced
• G3: Difficulty vs vanilla addressed
• G4: Compatibility/interactions checked

Gate Enforcement

enforcement_protocol = {
    "on_failure": "run_auto_fix",
    "max_retries": 2,
    "on_persistent_failure": "emit_limitation_notice",
    "continue": True  # Don't block entire output
}

Graceful Degradation

Degradation Levels

| Level | Condition | Behavior | |-------|-----------|----------| | 0 | All sources reachable | Full analysis | | 1 | Some primary fail | Use secondary + flag substitutions | | 2 | Most live fail | Knowledge base only + "historical" flag | | 3 | Variables missing | Proceed with available + "UNAVAILABLE" marks | | 4 | Complete failure | Emit notice, don't fabricate |

Limitation Banner

---
⚠️ LIMITATION NOTICE
This output was generated with reduced data availability (Level [0-4]).
Cross-check with current data before acting. Substituted/missing sources flagged inline.
---

Performance Optimization

Context Window Management

• Max tokens: 100,000 (configurable)
• Compression: Remove redundant context
• Caching: Cache knowledge base queries (1 hour TTL)
• Lazy loading: Load references only when needed

Token Optimization

• Summarization: Compress intermediate outputs
• Selective citation: Include only most relevant sources
• Template compression: Use compact output formats
• State pruning: Drop non-essential state data

Logging & Monitoring

Structured Logging

All events logged as JSON:

{
  "timestamp": "ISO-8601",
  "level": "INFO|WARNING|ERROR",
  "session_id": "string",
  "event": "hook_name|step_name",
  "data": {},
  "duration_ms": 0.0
}

Metrics Tracked

• Step execution times
• Gate pass/fail rates
• Degradation level changes
• Source availability
• Token consumption

Configuration

Required Files

• /config/settings.json - Main configuration
• /config/hooks.json - Hooks configuration
• /references/domain_knowledge.md - Domain references
• /references/balance_formulas.md - Mathematical formulas
• /SECOND-KNOWLEDGE-BRAIN.md - Living knowledge base

Environment Variables

Override config with:

export GAME_MOD_LOG_LEVEL=DEBUG
export GAME_MOD_MAX_WORKERS=8
export GAME_MOD_CACHE_TTL=7200

Dependencies

Required Tools

• WebSearch - Live data fetching
• WebFetch - Document retrieval
• Read - File reading
• Write - Knowledge base updates
• Bash - Script execution
• Skill - Sub-skill invocation

Optional Python Packages

requests>=2.28.0      # HTTP client
feedparser>=6.0.0     # RSS parsing
python-dateutil>=2.8.0  # Date parsing

Installation & Usage

Skill Installation

# Copy skill directory to Claude skills path
cp -r game-mod-balance-scaling ~/.claude/skills/

# Or install via package manager (if available)
claude skill install game-mod-balance-scaling

Basic Usage

User: Analyze the balance scaling for my weapon damage mod

[Skill activates → runs through harness → delivers report]

User: Can you check the economy consistency of this crafting mod?

[Skill activates → focuses on economy analysis → delivers report]

Testing & Validation

Test Scenarios

See /tests/test-scenarios.md for:

1. Standard analysis
2. Minimal input handling
3. Comparison scenarios
4. Risk/conflict detection
5. Degraded mode operation

Validation

Run project validation:

python scripts/validation.py --full

Maintenance

Knowledge Updates

Automatic via cron:

# Weekly academic update (Mondays 8:00 AM)
0 8 * * 1 python tools/knowledge_updater.py

# Daily news update (Daily 7:00 AM)
0 7 * * * python tools/knowledge_updater.py --news-only

Configuration Updates

1. Edit /config/settings.json
2. Run validation: python scripts/validation.py --config
3. Restart skill session

License & Attribution

License: MIT Author: 972026 Skill Library Version: 1.0.0 Last Updated: 2026-07-28

References

• Project documentation: /PROJECT-detail.md
• Development tracking: /PROJECT-DEVELOPMENT-PHASE-TRACKING.md
• Harness standard: /D:/972026/SKILL-STANDARD.md
• Reference implementation: /D:/vn-finance-analysis-hd-skill/

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: dungnotnull
• Source: dungnotnull/game-mod-balance-scaling-agent-skill
• License: MIT

Install and usage instructions live in the source repository linked above.