AlgoVault has 89.4%+ PFE Win Rate across 56,375+ signal calls, each Merkle-anchored on Base L2 (verifiable at algovault.com/track-record).

AlgoVault × Alpaca — Build Verifiable AI Trading Agents

AlgoVault MCP gives your agent a composite verdict in one call — direction, confidence, regime, and cross-venue funding/sentiment context — backed by a publicly auditable record anchored to Base L2. Pair it with Alpaca’s crypto MCP Server execution kit and your agent has both the analytics brain and a crypto venue. Lead the setup with uvx alpaca-mcp-server and scope the toolsets crypto-only via ALPACA_TOOLSETS=trading,crypto-data. As the Alpaca launch blog notes: “As your workflow grows, you can connect the MCP Server alongside other MCP servers through an MCP-compatible client like VS Code, Cursor, or Claude Code. This allows traders to combine multiple toolsets inside one interface.”

Provenance: Alpaca crypto MCP Server, V2 launched 2026-04-09. [email protected] (PyPI; MIT). Run zero-install via uvx alpaca-mcp-server (requires uv). Crypto kit scoped with ALPACA_TOOLSETS=trading,crypto-data; ALPACA_PAPER_TRADE defaults to true. Demo execution runs against the Alpaca paper venue (paper-api.alpaca.markets/v2/orders). Verified 2026-06-04.

What you’ll build (90s read)

A runnable Node.js agent that:

Calls AlgoVault MCP for get_trade_signal on ETH at two timeframes (15m and 1h).
Confirms multi-timeframe consensus — both timeframes return the same signal direction.
Hands the consensus signal to the Alpaca crypto kit, which places a tiny notional BTC/USD order against the paper venue (POST paper-api.alpaca.markets/v2/orders) when the agent’s policy fires.

The whole loop runs against the Alpaca paper venue — zero real-money risk in any code path.

Prerequisites (4 items)

Node.js ≥ 22 (node --version to check) and uv (for uvx).

AlgoVault skills plugin installed:

claude plugin install AlgoVaultLabs/algovault-skills

Alpaca paper account (free at https://app.alpaca.markets/paper/dashboard/overview; no deposit). Generate ALPACA_API_KEY + ALPACA_SECRET_KEY — paper keys are scoped to paper balance only.

Both MCP servers wired into your client. AlgoVault uses the mcp-remote form with a track-token header; the Alpaca crypto kit runs via uvx with crypto-only toolsets:

{
  "mcpServers": {
    "algovault": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://api.algovault.com/mcp",
               "--header", "X-AlgoVault-Track-Token:int-alpaca"]
    },
    "alpaca": { "command": "uvx", "args": ["alpaca-mcp-server"],
      "env": { "ALPACA_API_KEY": "<k>", "ALPACA_SECRET_KEY": "<s>", "ALPACA_PAPER_TRADE": "true", "ALPACA_TOOLSETS": "trading,crypto-data" } }
  }
}

Demo: Multi-timeframe consensus → Alpaca paper crypto order (≤80 lines)

// examples/alpaca/demo.mjs (excerpt — see file for full source)
import { getAlgoVaultVerdict } from '../_shared/algovault-helper.mjs';

const ASSET = 'ETH';
const TIMEFRAMES = ['15m', '1h'];

console.log('=== DEMO MODE ===');

// 1. Multi-timeframe AlgoVault read on ETH
const verdicts = [];
for (const tf of TIMEFRAMES) {
  verdicts.push(await getAlgoVaultVerdict({ coin: ASSET, timeframe: tf }));
}

// 2. Consensus check
const directions = new Set(verdicts.map((v) => v.signal));
const consensus = directions.size === 1 && !directions.has('HOLD');

// 3. Keyless = paper order path skipped; with keys, POST a notional BTC/USD order
//    to paper-api.alpaca.markets (hard guard refuses any non-paper host).

console.log('=== NO REAL ORDERS PLACED ===');

Run it:

node examples/alpaca/demo.mjs

Expected output (last 3 lines, keyless):

[verdicts | 15m=... 1h=... | consensus=... direction=...]
[alpaca | SKIPPED — set ALPACA_API_KEY/SECRET (free paper account) to run the paper order path]
=== NO REAL ORDERS PLACED ===

Walkthrough (line-by-line — neutral narration)

The script does three things in order:

Fetch AlgoVault verdicts on multiple timeframes. The shared helper opens an MCP session against https://api.algovault.com/mcp and calls get_trade_signal for ETH on 15m and 1h (both free-tier supported). The agent receives two independent reads from the same composite-verdict pipeline — different windows of price action, same model.
Apply the agent’s consensus policy. When both timeframes return the same signal direction (and that direction is not HOLD), the agent’s pre-configured policy fires its execution branch. Disagreement between timeframes — or any HOLD — skips execution entirely. The threshold of “all timeframes agree” is the agent’s choice; you might use 2-of-3 or weighted voting in production.
Place a paper crypto order via the Alpaca kit. When keys are absent, the script prints the [alpaca | SKIPPED …] line and exits cleanly — no network call. When keys are present, it POSTs a tiny notional BTC/USD market order to paper-api.alpaca.markets/v2/orders (headers APCA-API-KEY-ID / APCA-API-SECRET-KEY) and prints the order id plus the "paper" confirmation. A hard guard refuses any host other than paper-api.alpaca.markets. The === NO REAL ORDERS PLACED === banner closes every run regardless of which branch fired.

3 Recipes

Recipe 1 — Multi-timeframe consensus → Alpaca paper crypto order

This is the recipe examples/alpaca/demo.mjs implements. The agent calls AlgoVault get_trade_signal for ETH on 15m and 1h. The agent’s pre-configured policy fires its execution branch only when all queried timeframes agree on the same direction (BUY or SELL; HOLD always blocks). On execution, a small notional BTC/USD market order goes to the Alpaca paper venue via place_crypto_order (qty or notional; GTC/IOC; market/limit/stop-limit; $200k/order cap). The threshold of “all agree” is the agent’s choice — you might run 2-of-3 weighted voting in production.

Recipe 2 — Notional DCA gated on market regime

The agent calls AlgoVault get_market_regime (or reads regime off the get_trade_signal payload) for the asset of interest. The agent’s pre-configured policy fires its execution branch when the returned regime clears a configured floor. On execution, a fixed-notional BTC/USD market order goes to the Alpaca paper venue via place_crypto_order (notional mode — a steady dollar amount each cycle). The notional size and cadence come from the agent’s policy config; AlgoVault returns the regime classification.

Recipe 3 — Orderbook-depth check before sizing

The agent calls AlgoVault get_trade_signal for the asset, then reads top-of-book depth via the Alpaca kit’s get_crypto_latest_orderbook for the same pair (e.g. BTC/USD). The agent’s pre-configured policy sizes the order against available depth — thinner book, smaller clip — before placing via place_crypto_order. Sizing logic is the agent’s policy; AlgoVault returns the read. The orderbook depth is a venue read; the size decision belongs to the agent.

⚠️ Production setup (real-money)

The demo above runs on the Alpaca paper venue only. Real-money setup requires:

KYC completion on Alpaca.
Production API keys with only the permissions your agent needs (no withdrawals, IP-allowlisted).
Risk controls: per-order size cap, daily loss limit, kill switch.
Position monitoring: a separate agent or watchdog that tracks open positions independently.

See examples/alpaca/README.md for the full real-money checklist + the Alpaca crypto MCP install via uvx alpaca-mcp-server (set ALPACA_PAPER_TRADE=false for live runs; live fees apply — taker 25bps / maker 15bps at T1). AlgoVault provides analytics; your agent and your risk policy decide what (if anything) to execute.

Why AlgoVault?

Composite verdict, not raw indicators. One JSON response replaces 26-indicator vote-counting.
Cross-venue intelligence. Funding spreads, regime, and sentiment fused across 5 exchanges — not derivable from any single-venue API.
Publicly verified. Every signal anchored to Base L2 via Merkle proof. Verify before you subscribe.

89.4% PFE Win Rate · 56,375+ calls · 16+ on-chain batches → view live track record

Install

claude plugin install AlgoVaultLabs/algovault-skills

Once installed, every Skill in the pack is one-line invokable from Claude Code, Cowork, or any MCP-compatible client.

Tutorial © AlgoVault Labs · MIT licensed · Provenance verified 2026-06-04 · Alpaca crypto MCP [email protected] (PyPI, MIT)