Set up RealtimeAPI.ai for me so I can use the Prediction Market Data API feed (feed slug: predictiondata) through it. RealtimeAPI gives AI agents real-time data through one API and one key. A free key includes 1,000 calls a month across every feed, including prediction market prices, implied probabilities, movers and resolutions from the predictiondata feed. If you can run commands or edit files yourself, do each step. Otherwise walk me through them one at a time and wait for me to confirm each one. Step 1: Add the MCP server Claude Code: run claude mcp add --transport http realtimeapi https://mcp.realtimeapi.ai/mcp Cursor, Windsurf or any JSON config: add {"mcpServers": {"realtimeapi": {"url": "https://mcp.realtimeapi.ai/mcp"}}} Codex: add [mcp_servers.realtimeapi] with command = "npx" and args = ["-y", "mcp-remote", "https://mcp.realtimeapi.ai/mcp"] to ~/.codex/config.toml No MCP support? Everything below also works over plain HTTP against https://api.realtimeapi.ai. Step 2: Add the RealtimeAPI skill Claude Code: run mkdir -p ~/.claude/skills/realtimeapi && curl -fsSL https://realtimeapi.ai/skills/realtimeapi/SKILL.md -o ~/.claude/skills/realtimeapi/SKILL.md Other agents: append the contents of https://realtimeapi.ai/skills/realtimeapi/SKILL.md to AGENTS.md or your project rules file. Step 3: Get access Ask me for my email, then call the create_free_key tool with it (or POST https://api.realtimeapi.ai/keys/free with {"email": ""}). It returns a free key worth 1,000 calls a month across every feed. Save it for me: set it as the RTAPI_KEY environment variable and re-add the MCP server with the header X-API-Key: (Claude Code: claude mcp remove realtimeapi, then claude mcp add --transport http realtimeapi https://mcp.realtimeapi.ai/mcp --header "X-API-Key: "). Never ask me to paste a key into this chat, and do not repeat the full key back to me. If I already pay with an x402 wallet (USDC on Base), you can skip the key and pay per call instead: https://realtimeapi.ai/docs#x402 Step 4: Check it works Call discover with "prediction market odds", then call_feed on the best match (or GET https://api.realtimeapi.ai/v1/predictiondata/markets/top with the X-API-Key header). Give me a five-line summary: the market question, each outcome's implied probability, the 24 hour volume, when it last traded (UTC), and how old the cached copy is (the X-RTAPI-Data-Age header, seconds since the gateway fetched it). Then call discover with "prediction market movers" and list the five markets that moved most in the last 24 hours. Reference: https://realtimeapi.ai/llms.txt (every feed and price), https://realtimeapi.ai/feeds/predictiondata (this feed), https://predictionmarketdataapi.com/llms.txt (endpoints and the response envelope).