MidasFlow
AI
EN
MidasFlow
AI
Six ready-made skills that turn any Claude- or MCP-compatible agent into a disciplined market analyst on MidasFlow data: order flow, derivatives, liquidations, market phase and a calibrated P(TP1) across 8 exchanges. Install in one command.
Each skill is a playbook: when to fire, which grouped MCP tools to call and in what order, how to read the outputs, and how to synthesize one data-grounded verdict — never a trade signal.
1 · Connect the data (MCP)
claude mcp add --transport http midasflow https://mcp.midasflow.ai/mcp --header "Authorization: Bearer YOUR_KEY"2 · Add the skills
npx skills add midasflowai-lab/midasflow-skills| Skill | What it does | Key MCP tools |
|---|---|---|
midasflow-market-read | End-to-end situational read of a coin or the market — the default composition layer. | get_context · get_market · get_flow · get_whales · get_heatmap · score_symbol · get_accuracy · calc_ev |
midasflow-flow-interpretation | Read order flow and the book — CVD, VPIN, sweeps, walls, cross-exchange lead/lag. | get_flow · get_orderbook · get_whales |
midasflow-ev-discipline | Expected-value & honest-stats core: Wilson win-rate, break-even, EV. Catches "high-probability, losing bet". | get_accuracy · calc_ev · score_symbol · backtest |
midasflow-pump-read | Interpret one pump-family signal: real vs noise, early vs late/exhaustion, flow confirmation. | get_signals · get_context · score_symbol · get_flow |
midasflow-squeeze-setup | Read a squeeze setup from OI build-up, funding skew and liquidation clustering. | get_context · get_heatmap · get_whales · score_symbol |
midasflow-cross-exchange-leader | Which venue leads the tape; organic move (one leader) vs noise (all at once). | get_flow · get_orderbook · get_context |
Full MCP tool surface and parameters: API reference. Source & license (MIT): github.com/midasflowai-lab/midasflow-skills.