@undesirables/plugin-tcg-oracle

Give any ElizaOS agent real trading card market intelligence — search 370K+ products, grade cards with AI vision, and simulate future prices with Monte Carlo models.

Why This Exists

The trading card market is worth $50 billion and growing. Millions of people buy, sell, and grade cards every day. But the tools available to them are fragmented: you check prices on one site, send cards to PSA for $50+ and wait weeks for a grade, and have no way to forecast whether a card is going up or down.

This plugin gives your ElizaOS agent the ability to do all of that in one conversation:

1. "Search for Charizard cards" → Instantly searches 370K+ real products across 25 games
2. "Grade this card" → AI vision analyzes centering, corners, edges, surface → predicts PSA/Beckett score
3. "Simulate Charizard VMAX at $350 for 60 days" → Runs 10,000 Monte Carlo simulations using Heston, Merton, or Kou stochastic models → returns percentile price bands
4. "Show me the Pokemon market" → Pulls the latest daily market snapshot with top movers and volume trends

No API keys required from third parties. You just point it at the TCG Oracle server (self-hosted or public).

Quick Start

npm install @undesirables/plugin-tcg-oracle

Add to your ElizaOS character file:

{
  "plugins": ["@undesirables/plugin-tcg-oracle"],
  "settings": {
    "TCG_ORACLE_URL": "https://oracle.the-undesirables.com"
  }
}

Start your agent:

elizaos start --character your-character.json

That's it. Your agent can now search cards, grade images, and run simulations.

Actions

| Action | What It Does | Example Prompt | |--------|-------------|----------------| | TCG_ORACLE_SEARCH | Search 370K+ products by card name, set, or keyword | "Search for Base Set Charizard" | | TCG_ORACLE_GRADE | Grade a card image URL — returns PSA/Beckett prediction | "Grade this card: https://..." | | TCG_ORACLE_SIMULATE | Monte Carlo price forecast with 3 stochastic models | "Simulate Charizard at $350 for 60 days" | | TCG_ORACLE_MARKET | Pull the latest daily market snapshot for any game | "Show me the Pokemon market" |

Supported Games (25)

Pokémon · Magic: The Gathering · Yu-Gi-Oh! · Disney Lorcana · One Piece · Flesh and Blood · Star Wars: Unlimited · Dragon Ball Super · Digimon · MetaZoo · Union Arena · Gundam Card Game · LoL Riftbound · and 12 more via TCGCSV

How the Data Works

All market data comes from TCGCSV — a community project that snapshots pricing data daily for 25 TCG games. Our GitLab CI pipeline automatically refreshes this data every 24 hours. The full dataset (370K+ products) is also available on Kaggle.

Monte Carlo models explained:

• Heston — Models volatility itself as stochastic (volatility changes over time). Best for cards with unpredictable hype cycles.
• Merton — Adds random jumps to the price process. Best for cards that can spike on tournament results or reprints.
• Kou — Uses double-exponential jump distribution. Best for modeling asymmetric risk (big upside spikes vs gradual decline).

Self-Hosting

You can run your own TCG Oracle server instead of using the public API:

pip install undesirables-mcp-server
undesirables-mcp --workspace ./my-workspace

Then point TCG_ORACLE_URL to http://localhost:8000.

Configuration

| Setting | Description | Required | |---------|-------------|----------| | TCG_ORACLE_URL | Base URL of the TCG Oracle API | Yes |

Architecture

This is a native TypeScript ElizaOS plugin. It runs natively in the ElizaOS Node.js runtime — no Python subprocess, no external CLI tools. It calls the TCG Oracle REST API via fetch() with 30-second timeouts and structured parameters for v2 LLM parameter extraction.

Related Projects

• undesirables-mcp-server — 36-tool MCP server (Glama Registry, PyPI)
• plugin-undesirables — Soul personality plugin for ElizaOS
• tcg-oracle-tools — Python SDK for TCG market analysis

License

BUSL-1.1 — Business Source License 1.1. Copyright © 2026 The Undesirables LLC.