> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.soul.mds.markets/soul-hunt/ways-to-play/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.soul.mds.markets/_mcp/server. # Ways to play Pick by what you need. To predict and follow signals, use the Base App or soulhunt.ai. For investigation tools, use the MCP Server or the SDK. The Telegram bot does both and bundles a wallet. ## Base App Soul Hunt is a [Base App](https://www.base.org/build/base-app). Open [soulhunt.ai](https://soulhunt.ai) inside the Base App on your phone for one-tap sign-in with your Base wallet, with no email, seed phrase, or extra account. Stakes, payouts, and escrow settle in USDC on Base. | Feature | Available | | ----------------------------- | ---------------- | | One-tap sign-in (Base wallet) | Yes | | Submit predictions | Yes | | Subscribe to signals | Yes | | Strategist AI coach | Yes | | Instant USDC withdraw on Base | Yes | | Investigation tools | No (use MCP/SDK) | Best if you already have a Base wallet. ## Telegram Bot [@SoulMarketsBot](https://t.me/SoulMarketsBot) has a built-in wallet, real-time notifications when souls go live, and prediction submission and attempt tracking. | Feature | Available | | ------------------------- | --------- | | Submit predictions | Yes | | Real-time notifications | Yes | | Built-in USDC wallet | Yes | | Investigation tools | Yes | | Track attempts and scores | Yes | Start here if you're new and want a wallet bundled in. ## MCP Server The OneShot MCP server puts the full set of investigation tools in Claude Desktop or Cursor: research people, read traces, analyze patterns, and browse the web. | Feature | Available | | ----------------------- | ------------------- | | All investigation tools | Yes | | AI-assisted analysis | Yes | | Submit predictions | Via Telegram or API | | Real-time notifications | No | Research in Claude, then submit predictions through Telegram or the API. ```json { "mcpServers": { "oneshot": { "command": "npx", "args": ["-y", "@oneshot-agent/mcp-server"], "env": { "ONESHOT_WALLET_PRIVATE_KEY": "0x..." } } } } ``` ## OneShot SDK The SDK gives programmatic access to everything, for developers who want to automate: build your own investigation agent, auto-research souls, batch-analyze traces. | Feature | Available | | ----------------------- | --------- | | All investigation tools | Yes | | Programmatic access | Yes | | Build custom agents | Yes | | Submit predictions | Via API | **`TypeScript`** ```typescript TypeScript import { OneShot } from "@oneshot-agent/sdk"; const agent = new OneShot({ privateKey: process.env.AGENT_PRIVATE_KEY!, }); const research = await agent.research({ topic: "vitalik buterin recent DeFi opinions", depth: "deep", }); ``` **`Python`** ```python Python from oneshot import OneShotClient client = OneShotClient(private_key="0x...") result = client.call_tool("research", { "topic": "vitalik buterin recent DeFi opinions", "depth": "deep", }) ``` ## ChatGPT or any LLM You can paste traces into ChatGPT and reason about behavior, but you can't pull traces directly, use enrichment tools, or submit predictions. | Feature | Available | | ---------------------- | ---------------- | | Reasoning about traces | Yes (copy-paste) | | Direct trace access | No | | Investigation tools | No | | Submit predictions | No | ## Web (soulhunt.ai) The web app lets you browse souls, submit predictions, subscribe to signal feeds, and use the Strategist AI coach. | Feature | Available | | ---------------------------- | ---------------- | | Browse souls | Yes | | Read public traces | Yes | | Submit predictions | Yes | | Subscribe to signals | Yes | | Strategist AI coach | Yes | | Deposit USDC (card + crypto) | Yes | | Investigation tools | No (use MCP/SDK) | Use the web for predicting and subscribing, and the MCP Server or SDK for investigation tools. ## Playing outside the ecosystem Playing without OneShot tools caps your score. 2-5 USDC of tool spending moves you from a blind 30 to a realistic 80+. Without the tools you don't get: * Structured traces (just public summaries) * X clue trail analysis (@mysoulhunt) * Enrichment data from research tools * The spending-to-score correlation that makes the funnel work Generic predictions score low however good the reasoning is. Specific predictions need the data the tools provide. > Six entry points, ranked by how much they let you do.