360 Secure SkillHub
360 安全技能中心

optionwhales

该 Skill 通过 OptionWhales API 获取实时与历史期权资金流、异常交易和动量排名,并调用 AI 报告服务生成包含市场、基本面、新闻、辩论和风险分析的交易报告。使用前需配置相应 API 密钥。

psdsam-niuniuking
psdsam-niuniuking
热度 64.51
最近更新 2026-05-11 17:28:15
分类:数据依赖:代码执行 / 第三方 API应用场景:市场分析 / 报告撰写 / 财务 / 研究 ...

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add optionwhales
Metadata
name
option_whales
description
Query real-time option flow intelligence and generate AI-powered trade analysis reports. Use when users ask about option flow, unusual options activity, institutional positioning, market sentiment from derivatives, or AI trade reports.
homepage
https://www.optionwhales.io
metadata
{"openclaw":{"emoji":"🐋","requires":{"anyBins":["python3","python"],"env":["OPTIONWHALES_API_KEY"]},"primaryEnv":"OPTIONWHALES_API_KEY"}}

OptionWhales Intelligence Skill

Two capabilities in one skill:

  1. Option Flow — query the OptionWhales Pro API for real-time institutional option flow analysis (intent momentum, abnormal trades, rankings).
  2. AI Reports — generate and retrieve AI-powered trade analysis reports via a 22-node LLM pipeline (market analysis, flow, fundamentals, news, bull/bear debate, risk assessment, BUY/SELL/HOLD recommendation).

Free tier available! Sign up at https://www.optionwhales.io to get a free API key instantly — no credit card required.

When to Use

✅ USE this skill when:

  • "What's the option flow on AAPL today?"
  • "Show me unusual options activity"
  • "What are institutions buying?"
  • "Is there bullish or bearish momentum in TSLA?"
  • "Show me the top momentum tickers"
  • "Generate an AI report for TSLA"
  • "What's the AI analysis on SPY today?"
  • "Show my report history"
  • "Check my report generation quota"

When NOT to Use

❌ DON'T use this skill when:

  • Stock price quotes only → use a stock/market data skill
  • Historical stock charts → use a charting skill
  • GEX / gamma exposure queries → not exposed via API
  • Options pricing or Greeks calculation → use a quant tool
  • Crypto or forex → this is US equity options only

Setup

Option Flow (required)

Requires an OptionWhales API key.

export OPTIONWHALES_API_KEY="ow_free_your_key_here"  # or ow_pro_...

AI Reports (optional)

Requires an AI service Bearer token (contact system administrator).

export AI_API_TOKEN="your_bearer_token_here"

Part 1 — Option Flow API

Base URL: https://api.optionwhales.io/v1

All requests require header: X-API-Key: $OPTIONWHALES_API_KEY

Flow API Reference

Current Flow Rankings

curl -s -H "X-API-Key: $OPTIONWHALES_API_KEY" \
  "https://api.optionwhales.io/v1/flow/current" | python3 -m json.tool

Response includes per-ticker: intent_primary (Directional/Gamma/LongVol/ShortVol/Mixed), direction_bias (bullish/bearish/neutral), intent_confidence (0-1), momentum_fast, momentum_slow, coherence_last, strength_last, key_strikes.

Ticker Detail (Clusters + Time Series)

curl -s -H "X-API-Key: $OPTIONWHALES_API_KEY" \
  "https://api.optionwhales.io/v1/flow/current/AAPL" | python3 -m json.tool

Returns: ranking (summary), clusters (strategy clusters with dollar Greeks), cluster_trades (individual trades), time_series (30-min momentum buckets).

Historical Sessions

# System health and session status
curl -s -H "X-API-Key: $OPTIONWHALES_API_KEY" \
  "https://api.optionwhales.io/v1/flow/sessions" | python3 -m json.tool

# Get a specific session's rankings
curl -s -H "X-API-Key: $OPTIONWHALES_API_KEY" \
  "https://api.optionwhales.io/v1/flow/2025-06-02" | python3 -m json.tool

# Get a specific ticker from a historical session
curl -s -H "X-API-Key: $OPTIONWHALES_API_KEY" \
  "https://api.optionwhales.io/v1/flow/2025-06-02/AAPL" | python3 -m json.tool

Momentum Rankings

curl -s -H "X-API-Key: $OPTIONWHALES_API_KEY" \
  "https://api.optionwhales.io/v1/momentum/rankings" | python3 -m json.tool

Momentum History

curl -s -H "X-API-Key: $OPTIONWHALES_API_KEY" \
  "https://api.optionwhales.io/v1/momentum/AAPL/history" | python3 -m json.tool

Abnormal Trades

# Current session
curl -s -H "X-API-Key: $OPTIONWHALES_API_KEY" \
  "https://api.optionwhales.io/v1/abnormal-trades/current" | python3 -m json.tool

# Historical session
curl -s -H "X-API-Key: $OPTIONWHALES_API_KEY" \
  "https://api.optionwhales.io/v1/abnormal-trades/2025-06-02" | python3 -m json.tool

Account Usage

curl -s -H "X-API-Key: $OPTIONWHALES_API_KEY" \
  "https://api.optionwhales.io/v1/account/usage" | python3 -m json.tool

WebSocket — Real-Time Abnormal Trades (Pro Only)

python3 -c "
import asyncio, websockets, json
async def stream():
    uri = 'wss://api.optionwhales.io/v1/ws/abnormal-trades?api_key=YOUR_PRO_KEY'
    async with websockets.connect(uri) as ws:
        await ws.send(json.dumps({'type': 'subscribe', 'tickers': ['AAPL', 'NVDA', 'TSLA']}))
        async for msg in ws:
            data = json.loads(msg)
            if data['type'] == 'abnormal_trade':
                print(json.dumps(data['data'], indent=2))
asyncio.run(stream())
"

Flow Helper Script

# Current flow rankings
python3 {baseDir}/scripts/optionflow.py flow

# Flow for specific ticker
python3 {baseDir}/scripts/optionflow.py flow --ticker AAPL

# Historical session flow
python3 {baseDir}/scripts/optionflow.py flow --session 2025-06-02

# Momentum rankings
python3 {baseDir}/scripts/optionflow.py momentum

# Top 5 momentum tickers
python3 {baseDir}/scripts/optionflow.py momentum --top 5

# Current abnormal trades
python3 {baseDir}/scripts/optionflow.py abnormal

# Historical abnormal trades
python3 {baseDir}/scripts/optionflow.py abnormal --session 2025-06-02

Interpreting Flow Results

Intent Types

Intent Meaning
Directional Net delta-dominant flow — strong directional bet
Gamma Gamma-dominant — positioning for rapid price moves
LongVol Buying volatility (long vega, positive theta risk)
ShortVol Selling volatility (short vega, positive theta)
Mixed No dominant Greek — conflicting signals

Direction Bias

  • bullish: Net positive delta flow
  • bearish: Net negative delta flow
  • neutral: Delta share below 15% of total flow

Confidence Score

0–1 scale combining flow coherence (how aligned the trades are) and strength (total dollar Greek magnitude). Above 0.7 is high conviction.

Momentum

  • momentum_fast (α=0.35): responsive to recent flow changes
  • momentum_slow (α=0.15): trend-smoothed
  • Positive = bullish acceleration, negative = bearish acceleration

Part 2 — AI Report Service

Base URL: https://ai-service-production-b44b.up.railway.app

All requests require header: Authorization: Bearer $AI_API_TOKEN

Credit System

Tier Credits/Day Can Generate?
FREE 0 No — hard-blocked
TRIAL 5 Yes
PRO 5 Yes
ADMIN Unlimited Yes

Additional constraints:

  • 5-minute cooldown per ticker after a successful generation
  • ≥10% premium increase required to regenerate for the same ticker
  • Credits reset daily at 12:00 AM ET

AI Report API Reference

Generate a Report

curl -s -X POST \
  -H "Authorization: Bearer $AI_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"user_id":"user@example.com","user_tier":"PRO","ticker":"TSLA","session_date":"2026-03-25","session_type":"live","large_orders":[]}' \
  "https://ai-service-production-b44b.up.railway.app/reports"

Returns: {"job_id": "abc-123", "status": "queued", "message": "..."}

Poll Job Status

curl -s -H "Authorization: Bearer $AI_API_TOKEN" \
  "https://ai-service-production-b44b.up.railway.app/reports/{job_id}"

Fetch Report (JSON)

curl -s -H "Authorization: Bearer $AI_API_TOKEN" \
  "https://ai-service-production-b44b.up.railway.app/reports/by-id?user_id=user@example.com&ticker=TSLA&trading_day=2026-03-25&report_id=rpt-456"

Fetch Report (Markdown)

curl -s -H "Authorization: Bearer $AI_API_TOKEN" \
  "https://ai-service-production-b44b.up.railway.app/reports/{job_id}/artifact/md"

Report History

curl -s -H "Authorization: Bearer $AI_API_TOKEN" \
  "https://ai-service-production-b44b.up.railway.app/reports/history?ticker=TSLA&trading_day=2026-03-25&user_id=user@example.com"

Per-Ticker Summary

curl -s -H "Authorization: Bearer $AI_API_TOKEN" \
  "https://ai-service-production-b44b.up.railway.app/reports/history/summary?user_id=user@example.com&trading_day=2026-03-25"

Check Quota

curl -s -H "Authorization: Bearer $AI_API_TOKEN" \
  "https://ai-service-production-b44b.up.railway.app/quotas?user_id=user@example.com&tier=PRO"

Check Eligibility

curl -s -X POST \
  -H "Authorization: Bearer $AI_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"user_id":"user@example.com","user_tier":"PRO","ticker":"TSLA","session_date":"2026-03-25","session_type":"live","large_orders":[]}' \
  "https://ai-service-production-b44b.up.railway.app/eligibility"

Health Check

curl -s -H "Authorization: Bearer $AI_API_TOKEN" \
  "https://ai-service-production-b44b.up.railway.app/health"

AI Report Helper Script

# Generate a report
python3 {baseDir}/scripts/aireport.py generate --ticker TSLA --user-id user@example.com --user-tier PRO

# Check job status
python3 {baseDir}/scripts/aireport.py status --job-id abc-123

# Fetch report (JSON)
python3 {baseDir}/scripts/aireport.py report --user-id user@example.com --ticker TSLA --trading-day 2026-03-25 --report-id rpt-456

# Fetch report (Markdown)
python3 {baseDir}/scripts/aireport.py report --job-id abc-123 --format markdown

# Browse existing reports
python3 {baseDir}/scripts/aireport.py summary --user-id user@example.com --trading-day 2026-03-25

# List reports for a ticker
python3 {baseDir}/scripts/aireport.py history --ticker TSLA --trading-day 2026-03-25 --user-id user@example.com

# Check quota
python3 {baseDir}/scripts/aireport.py quota --user-id user@example.com --tier PRO

# Check eligibility
python3 {baseDir}/scripts/aireport.py eligibility --ticker TSLA --user-id user@example.com --user-tier PRO

# Service health
python3 {baseDir}/scripts/aireport.py health

Interpreting AI Report Results

Decision Actions

Action Meaning
BUY Bullish — AI recommends long exposure on the underlying
SELL Bearish — AI recommends short exposure / protective puts
HOLD Neutral — no new directional exposure recommended

Confidence Score

0–100 scale reflecting the AI pipeline's conviction. Above 70% = high conviction; below 50% = conflicting signals.

Report Sections

Section Content
Option Flow Institutional flow analysis, intent momentum, flow coherence
Abnormal Flow Large/unusual trades flagged by the detection engine
Market Structure GEX levels, call/put walls, max pain, key strike zones
Market Macro indicators, OHLCV, technical levels
Fundamentals Company fundamentals, valuation metrics
News Recent news sentiment analysis
Debate Bull vs Bear researcher arguments with judge decision
Risk Three-analyst risk assessment (aggressive, conservative, neutral)

Example Queries → Commands

User asks Tool
"What's the flow today?" python3 {baseDir}/scripts/optionflow.py flow
"Show TSLA option activity" python3 {baseDir}/scripts/optionflow.py flow --ticker TSLA
"Top momentum tickers" python3 {baseDir}/scripts/optionflow.py momentum
"Any unusual trades?" python3 {baseDir}/scripts/optionflow.py abnormal
"Is AAPL bullish or bearish?" python3 {baseDir}/scripts/optionflow.py flow --ticker AAPL → check direction_bias
"Generate AI report for TSLA" python3 {baseDir}/scripts/aireport.py generate --ticker TSLA --user-id <user> --user-tier PRO
"Show my report history" python3 {baseDir}/scripts/aireport.py summary --user-id <user> --trading-day <today>
"Check my report quota" python3 {baseDir}/scripts/aireport.py quota --user-id <user> --tier PRO
"Show my API usage" python3 {baseDir}/scripts/optionflow.py usage

Notes

  • Option flow data updates every 30 minutes during market hours (9:30 AM – 4:00 PM ET)
  • Final daily data available after 7:30 AM ET next trading day (with full OI deltas)
  • Free tier: 10 requests/minute, 200/day; top 3 tickers, last 5 abnormal trades
  • Pro tier: 60 requests/minute, 5000/day; all tickers, all fields, WebSocket, historical sessions
  • AI reports take 30–90 seconds to generate (22-node pipeline)
  • US equity options only (no futures, crypto, or forex)