#OpenAlgo Execution Skills for Agentic Coding Tools

A skill pack for building production-grade algorithmic trading strategies on the OpenAlgo platform. Every strategy generated by these skills is a single Python file that toggles between backtest mode (VectorBT) and live execution mode (OpenAlgo SDK + WebSocket) via one flag — no separate codebases.

Strategies are also upload-ready for OpenAlgo's self-hosted /python strategy host with exchange-aware scheduling, env-var driven configuration, and SIGTERM-safe shutdown.

Works with 40+ AI coding agents via skills.sh — Claude Code, Cursor, Codex, OpenCode, Cline, Windsurf, GitHub Copilot, Gemini CLI, Roo Code, and more.

#Quick Install

# GitHub shorthand
npx skills add marketcalls/openalgo-execution-skills

# Specific skill
npx skills add marketcalls/openalgo-execution-skills -s algo-strategy

#Slash Commands

#Single-File Dual-Mode Pattern

Every generated strategy looks like this:

# Local backtest (uses VectorBT)
python strategies/my_ema/strategy.py --mode backtest

# Local live (real OpenAlgo orders + WS risk manager)
# Live vs sandbox is controlled in OpenAlgo's UI analyzer toggle
python strategies/my_ema/strategy.py --mode live

# Self-hosted via OpenAlgo /python (env-driven, no CLI)
# Upload through http://localhost:5000/python

The same signals(df) function feeds both VectorBT (backtest) and the live event loop. The same risk thresholds (SL/TP/trailing) apply in both modes. The same fee and slippage assumptions are honored on both sides.

#Three Execution Types

Asked at strategy creation time, baked into the file:

#Indicator Library Choice

Asked at strategy creation. Default is openalgo.ta (Numba-JIT, 100+ indicators). User can pick talib for standard indicators only — specialty indicators (Supertrend, Ichimoku, Donchian, HMA, KAMA) always come from openalgo since talib doesn't have them.

#Real-World Cost Modeling

4-segment Indian market fees baked in (matches vectorbt-backtesting-skills conventions):

Backtest applies these via VectorBT fees, fixed_fees, slippage parameters. Live mode optionally uses LIMIT-with-offset to control slippage and tracks measured slippage per fill, with end-of-session drift report.

#Risk Management

Per-position (in every strategy):

• Stop loss (% or absolute)
• Take profit (% or absolute)
• Trailing stop (% or absolute, watermark-based)
• Time-based exit (max hold minutes)

Portfolio-level (via /algo-portfolio):

• Portfolio SL / TP (e.g. -2% / +3% of capital)
• Daily PnL stop (resets at IST midnight)
• Daily PnL target
• Max concurrent positions
• Max symbol concentration

All caps work in both backtest and live modes — backtest computes equity per bar inside the runner, live tracks realized + unrealized P&L from tradebook() and positionbook().

#Self-Hosted via OpenAlgo /python

Every strategy is upload-ready for OpenAlgo's built-in /python strategy host:

• Reads OPENALGO_STRATEGY_EXCHANGE, HOST_SERVER, OPENALGO_API_KEY from env (with fallbacks)
• SIGTERM-safe shutdown (graceful WS disconnect, state flush)
• stdout-only logging (host captures to logs/strategies/)
• SQLite state per strategy survives restarts
• Exchange-aware calendar gating works automatically

The /algo-host skill validates compatibility and generates an upload checklist with exact form values to enter on http://localhost:5000/python.

#Strategy Templates (12)

#Prerequisites

#1. OpenAlgo Platform

git clone https://github.com/marketcalls/openalgo.git
cd openalgo
pip install -r requirements.txt
python app.py

OpenAlgo runs at http://127.0.0.1:5000. WebSocket at ws://127.0.0.1:8765. Connect a broker via the OpenAlgo dashboard and grab the API key.

#2. Python Environment

python -m venv venv
source venv/bin/activate           # Linux/Mac
# venv\Scripts\activate            # Windows

pip install -r requirements.txt

#3. Configure API Keys

cp .env.sample .env
# Edit .env with OPENALGO_API_KEY and host URLs

#Configuration

The .env file is read by every generated strategy:

OPENALGO_API_KEY=your_api_key_here
HOST_SERVER=http://127.0.0.1:5000
WEBSOCKET_URL=ws://127.0.0.1:8765

When uploaded to OpenAlgo's /python, the platform injects OPENALGO_STRATEGY_EXCHANGE, OPENALGO_API_KEY, STRATEGY_ID, STRATEGY_NAME and inherits HOST_SERVER and WEBSOCKET_URL from OpenAlgo's own .env.

#Knowledge Base (23 Rule Files)

#Companion Skill Packs

This pack composes naturally with:

• openalgo-indicator-skills — charting, dashboards, scanners, custom indicators
• vectorbt-backtesting-skills — pure backtesting (no live trading)

This pack is the execution + dual-mode layer. The other two are research / analysis layers.

#License

MIT