skill

game-lore-theorycrafting-mystery-agent-skill

📜🔍 Game Lore Theorycrafting & Hidden Mystery Decoding Engine | Production-grade Claude Code harness (v3.0.0) & lore_agent Python framework for Elden Ring, Dark Souls, ARG & Secrets. Features Tier 1-

dungnotnull/game-lore-theorycrafting-mystery-agent-skillcommunitylistedunreviewed

Use skill

Set up

Skills are installed by placing their folder where your agent loads skills.

Download ZIP from GitHub Open the repository

The download comes from github.com/dungnotnull/game-lore-theorycrafting-mystery-agent-skill itself, not from our server, so what you get is exactly what its author published.

Install instructions come from the README

Skills do not use an MCP config. Follow the steps in the README below.

README

synced Oct 22, 2026

game-lore-theorycrafting-mystery

Game Lore Theorycrafting & Hidden-Mystery Decoding

A production-grade Claude Code harness for Narrative Theorycrafting & Game-Lore Archaeology — gathers real-time authoritative data, applies recognized domain methods, integrates academic research, manages context budgets, and delivers evidence-backed, risk-disclosed outputs.

Features

Feature	Description
Real-time Analysis	Multi-source evidence aggregation (wikis, dev interviews, ARG communities, datamining references)
Systematic Methodology	5-step harness with 10 quality gates (U1-U6 + G1-G4) and auto-fix logic
Academic Integration	Self-updating knowledge base with Tier 1-4 evidence hierarchy and DOI-cited references
Context Management	Token budget enforcement with step-level allocation, evidence compression, and summary-only mode
Risk-Disclosed Outputs	Mandatory disclosure before conclusions, multi-scenario (best/base/worst), 5-verdict system
Self-Improving Pipeline	Automated weekly academic crawl + daily news crawl via ArXiv, Semantic Scholar, and RSS
Bilingual Support	English and Vietnamese output with automatic language detection
Graceful Degradation	5-level fallback chain (Level 0-4) with explicit LIMITATION banners

Architecture

USER INPUT
    │
    ▼
[Step 1: sub-gather-requirements]  → Clarify object, scope, language
    │
[Step 2: sub-evidence-collector]   → Fetch live data + authoritative docs
    │
[Step 3: sub-core-analysis]        → Lore theorycraft + ARG decode + canon/speculation
    │
[Step 4: sub-knowledge-updater]    → Academic evidence with Tier labels
    │
[Step 5: sub-advisor]              → Risk-disclosed conclusion + evidence chain
    │
[Step 6: Quality Gate Review]      → U1-U6 + G1-G4 auto-fix + enforcement
    │
    ▼
FINAL REPORT (with token budget tracking)

Flexible Agent Framework (v3.0.0)

In addition to the markdown skills, the harness ships a production-grade Python framework in lore_agent/ implementing a modular skill-registry pattern:

SkillRegistry — register / resolve / execute / validate skills (JSON schemas)
ChainOfThoughtRouter — rule + LLM routing with graceful fallback
BaseAgent + 5 sub-agents — bridge the LLM and structured JSON I/O
HookBus — lifecycle / state-sync / event-emission hooks
ContextManager — token budget, evidence compression, summary-only mode
ToolRegistry — web_search, web_fetch, knowledge_query, knowledge_append, cipher_decode (8 ciphers), language_detect
Provider-agnostic LLM layer (MockLLMClient offline, AnthropicLLMClient, OpenAILLMClient) with retry + graceful fallback
Type-safe config/ (Settings + FeatureFlags, env + YAML precedence)

Run end-to-end (fully offline by default):

python -m scripts.run_agent "decode the Gwyn mystery in Dark Souls"
python -m scripts.run_agent --query "..." --json
python -m scripts.run_agent --list-skills
python -m scripts.dump_manifests      # emit assets/schemas/*.manifest.json

See SKILL.md for the full registry contract.

Installation

git clone https://github.com/game-lore-theorycrafting-mystery.git
cd 199-game-lore-theorycrafting-mystery
pip install -r requirements.txt

Install skill files to ~/.claude/skills/ or use via project CLAUDE.md.

Usage

In Claude Code:

/game-lore-theorycrafting-mystery [your query]

Manual knowledge pipeline updates:

python tools/knowledge_updater.py --dry-run        # Preview without writing
python tools/knowledge_updater.py --validate-config # Validate configuration
python tools/knowledge_updater.py                   # Full crawl (academic + news)
python tools/knowledge_updater.py --news-only       # RSS news only
python tools/knowledge_updater.py --log-level DEBUG # Verbose output

Testing

python tools/test_knowledge_updater.py    # 21 unit tests
python tools/run_test_scenarios.py        # Structural & content validation

Knowledge Base

SECOND-KNOWLEDGE-BRAIN.md is auto-updated weekly via tools/knowledge_updater.py.

The pipeline crawls:

ArXiv (cs.CL, cs.AI, cs.CY, cs.HC, cs.CV, cs.MM)
Semantic Scholar (keyword search)
RSS Feeds (Game Studies, Kotaku, IGN, RPS, Polygon, GamesIndustry.biz, PlayStation Blog)

Set SEMANTIC_SCHOLAR_API_KEY in your environment for higher rate limits.

Data Sources

Source	Type	Tier
Official game lore wikis (Fextralife, Fandom)	Community-authoritative	Tier 2-3
In-game item/text databases	Primary canon	Tier 1
Developer interviews & patch notes	Primary authoritative	Tier 1
Fan lore archives	Community reference	Tier 3
ARG/secret-solving communities	Community investigative	Tier 2-3
Datamining references	Out-of-canon	Tier 4
Academic papers (ACM, Elsevier, etc.)	Peer-reviewed	Tier 1-2

Quality Gates

U1-U6: Universal gates (source count, disclosure, evidence hierarchy, language, template, traceability)
G1-G4: Domain gates (in-game evidence linking, canon/speculation flagging, datamine caveats, community corroboration)

Each gate includes auto-fix logic with 2-retry enforcement. See skills/main.md for the full gate specification.

Roadmap

Phase 0: Architecture & research
Phase 1: 5 core sub-skills
Phase 2: Main harness + 10 quality gates + degradation
Phase 3: Knowledge pipeline + tests + cron
Phase 4: Testing & validation
Phase 5: Integration & polish (v1.0.0)
Phase 6: Production hardening & open-source (v2.0.0)
Phase 7: Flexible Agent & Skill Architecture - lore_agent framework (v3.0.0)

Contributing

See CONTRIBUTING.md for setup, coding standards, and PR process. All contributions must pass the test suite and structural validator.

Security

See SECURITY.md for our security policy and vulnerability reporting process.

License

MIT — see LICENSE.

Citation

@software{game-lore-theorycrafting-mystery,
  title = {game-lore-theorycrafting-mystery: Game Lore Theorycrafting & Hidden-Mystery Decoding},
  year = {2026},
  version = {2.0.0},
  license = {MIT},
  url = {https://github.com/game-lore-theorycrafting-mystery}
}

Why This Skill

Narrative Theorycrafting & Game-Lore Archaeology practitioners face fragmented data, inconsistent methodology, and tools that do not self-improve. This skill unifies authoritative real-time data, recognized domain methods, and a continuously-updated academic knowledge base into one evidence-backed, risk-disclosed, context-managed workflow.