x402-tasignal

TASignal: paid x402 technical analysis (RSI, MACD, EMA/SMA, Bollinger, ATR, ADX, stochastic, realized volatility and a descriptive bias score; risk metrics with VaR/CVaR, Sharpe/Sortino, drawdown and ATR position sizing; rule backtests on closed public candles; read-only analytics and simulation, fail-closed; not financial advice). Cloudflare Worker, Hono, x402 v2 + v1, Base / Polygon / Arbitrum One / Solana, remote and stdio MCP.

Browse source · x402-tasignal.tar.gz · x402-tasignal.bundle
commit b8a491c · b8a491c9e0d4216ca20c45aee862054f7b96a80d
Clone: git clone https://agentgate-source.pages.dev/x402-tasignal/x402-tasignal.git (branch main, commit b8a491c9e0d4216ca20c45aee862054f7b96a80d) or

curl -O https://agentgate-source.pages.dev/x402-tasignal/x402-tasignal.bundle && git clone x402-tasignal.bundle x402-tasignal

Home: this repo is hosted here (it was never on GitHub).

TASignal: technical indicators, risk metrics, position sizing and backtests for crypto trading agents (x402 API + MCP)

One pay-per-call API for the chart work a trading agent does before it acts:

Indicators on closed 15m / 1h / 4h / 1d candles: RSI 14, MACD 12/26/9 (with cross), EMA 20/50/200, SMA 20, Bollinger Bands 20/2 (%b, bandwidth), ATR 14 and ATR %, ADX 14 (+DI/-DI), stochastic %K, annualized realized volatility, volume vs 20-bar average, plus a descriptive technical bias score (-100..100) with components, trend and regime.
Risk and sizing: historical VaR / CVaR 95/99 (per bar and per day), volatility, max and current drawdown, Sharpe / Sortino, ATR(14) stop or your stop, units, USD notional, implied leverage, 2R / 3R levels.
Backtests: sma_cross, ema_cross, rsi_reversion, breakout (Donchian), macd_cross on up to 2000 candles, long-only or long/short, fees per side, signals on close and fills at the next open. Return vs buy-and-hold, max drawdown, Sharpe, win rate, profit factor, exposure, the last 20 trades and an equity curve.

Not financial advice. Past performance does not guarantee future results. Indicators, risk metrics and backtests describe historical prices only. TASignal is read-only: it never trades, never holds funds and never asks for keys.

API: https://api.youraitools.co/tasignal (fallback https://tasignal.agentgate.workers.dev)
Remote MCP: https://api.youraitools.co/tasignal/mcp (tool ta_bias, free preview)
stdio MCP: npx -y @agentgate-builder/tasignal (tools ta_indicators, ta_position_size, ta_backtest)
OpenAPI: /openapi.json · x402 manifest: /.well-known/x402

Routes and prices

Route	Price	Example
`/v1/lite`	$0.005	`?symbol=BTC&timeframe=1h`
`/v1/risk`	$0.02	`?symbol=ETH&timeframe=4h&account_usd=10000&risk_pct=1&side=long`
`/v1/backtest`	$0.05	`?symbol=SOL&timeframe=4h&strategy=ema_cross&fast=20&slow=50`
`/v1/preview`	free (10/min/IP)	`?symbol=BTC&timeframe=1h` (bias, trend, regime only)

All paid routes accept POST with the same fields as JSON. Payment: x402 v2 (PAYMENT-SIGNATURE; USDC on Base, Polygon, Arbitrum One or Solana) or v1 (X-PAYMENT, Base), or a prepaid YourAITools credits key (Authorization: Bearer yatc_...). Bad input (400), unknown symbols (404) and upstream failures (503) are never charged.

Data and fail-closed behaviour

Candles come from Hyperliquid (candleSnapshot), with the dYdX v4 indexer as fallback; the source is named in every answer. If Hyperliquid fails, the answer says so in sources_failed and confidence drops. If no source answers: HTTP 503 upstream_unavailable, not charged. Only closed candles are used. An indicator that needs more history than is available is null with a note, never extrapolated.

stdio MCP paid mode

Set TASIGNAL_X402_PRIVATE_KEY (a Base wallet with a little USDC) and install the optional peers: npx -y -p @agentgate-builder/tasignal -p @x402/fetch -p @x402/evm -p viem tasignal. Spend guard: Base USDC only, max $0.05 per payment, only to the TASignal payTo.

Develop

npm test (offline mocks) · npm start (Node server) · wrangler deploy (Cloudflare Worker). MIT.

agentgate-source: code mirror on Cloudflare Pages. All repos