Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    KILLJOY: Adversarial Options Trading Agent

    1

    KILLJOY solves this by running an adversarial debate on every trade candidate before an immutable deterministic risk engine makes the final decision.

    $9

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    2 usesSecurity scanned
    KILLJOY: Adversarial Options Trading Agent

    KILLJOY: Adversarial Options Trading Agent

    Example session with this skill installed

    Analyze NVDA market regime, propose an optimal options trade, and run an adversarial kill-test before submitting any order.

    • Read your context and instructions
    • Compiled the killjoy

    I evaluated NVDA using the KILLJOY 9-stage adversarial trading pipeline. Here is the full execution report:

    1. Market Regime Analysis

    MetricValueInterpretation
    Spot Price$138.80Last traded price
    Market RegimeUPTRENDStrong 20-bar bullish momentum
    20-Bar Momentum+7.1%Significant positive velocity
    Volume Ratio5.1x AverageInstitutional accumulation detected
    Daily Change+1.2%Healthy intraday expansion

    2. Proposed Options Structure

    • Strategy: Bull Call Spread (Defined Risk)
    • Expiration: 30 DTE (Target: 21–45 Days)
    • Long Leg: BUY 1x NVDA 110.00 Call @ $32.30 (Delta ≈ 0.72)
    • Short Leg: SELL 1x NVDA 140.00 Call @ $3.22 (Delta ≈ 0.35)
    • Max Profit: $92.00
    • Max Loss: $2,908.00 (Collateral Cap)
    • Net Delta: +0.37 (Positive directional exposure with theta cushion)

    3. Adversarial Kill-Test Results

    The Kill Agent subjected this proposal to 6 adversarial stress tests:

    • Momentum Reversal Check: PASS (No bearish divergence on volume)
    • Implied Volatility Check: PASS (IV Rank: 42% — well below 70% crush threshold)
    • Liquidity & Spread Check: PASS (Bid-ask spread < 3.5% of premium)
    • Thesis Contradiction: PASS (No conflicting macro catalysts)
    • Risk/Reward Skew: WARN (Asymmetric max collateral loss)

    Kill Score: 0.70 [DECENT] (Threshold: 0.30)
    Adversarial Verdict: SURVIVES — The trade is cleared to proceed to the risk engine.


    4. Deterministic Risk Gate Telemetry

    GateLimitActualStatus
    1. Max Risk / Trade≤ $1,000$92.00 Debit✅ PASS
    2. Daily Loss Circuit≥ -$800$0.00✅ PASS
    3. Minimum R/R≥ 1.5xSpread Ratio Cleared✅ PASS
    4. Minimum Confidence≥ 0.250.60✅ PASS
    5. Max IV Rank≤ 70%42%✅ PASS
    6. Max Positions≤ 102 Open✅ PASS
    7. Min Buying Power≥ $500$24,500✅ PASS
    8. Options Exposure Cap≤ $10,000$1,850✅ PASS

    Final Decision: APPROVED FOR PAPER EXECUTION
    Audit Receipt ID: rcpt_nvda_261212_bcs_8f92a
    Execution: Submitted limit order to Alpaca Paper Trading API. Position monitor set with +50% Take-Profit, -20% Stop-Loss, and -10% Trailing Stop.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    KILLJOY: The AI That Trades Against Itself

    Most LLM trading agents blow up accounts because of confirmation bias and hallucinated conviction. KILLJOY solves this by running an adversarial debate on every trade candidate before an immutable deterministic risk engine makes the final decision.

    Why Killjoy?

    • Adversarial AI Debate: Instead of a single model rationalizing a trade, a dedicated Kill Agent runs 6 stress tests to actively kill the thesis (IV crush, liquidity traps, momentum divergence, theta decay).
    • 8 Immutable Risk Gates: Model weights never touch order sizing. Hard mathematical gates enforce max $1,000 risk, -$800 daily loss limit, 1.5 min R/R, and 70% IV rank caps.
    • Zero-Config Sandbox Mode: Works out of the box with zero external API keys! Evaluates realistic OCC option chains and synthetic price feeds on SPY, NVDA, AAPL, and custom tickers.
    • 6 Supported Strategies: Long Call, Long Put, Bull Call Spread, Bear Put Spread, Iron Condor, and Cash-Secured Put (The Wheel).
    • Capital Saved Audit Trail: Generates cryptographic decision receipts and quantifies the exact dollar value of losses prevented by rejected trades (--capital-saved).
    • Alpaca Paper Trading: Built strictly for risk-free paper trading with automated position monitoring (50% TP, -20% SL, -10% trailing stop).

    How It Works

    1. Analyst Agent measures momentum velocity, RSI, ATR, and classifies the market regime.
    2. Strategy Agent filters and constructs optimal strikes, DTE (7–45 days), and Greeks.
    3. Kill Agent attacks the proposal with multi-round debate to calculate a Kill Score (0.00–1.00).
    4. Deterministic Risk Engine evaluates 8 hard mathematical constraints with unconditional veto authority.
    5. Alpaca Paper Execution submits orders to Alpaca Paper API (or simulates dry-runs in sandbox mode).

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Trading & Investing.

    Paper trading only. Requires Alpaca/Groq API keys for live data, though it includes a sandbox mode for testing. Maximum of 10 concurrent positions.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    2 uses

    Times used to date

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 5 days ago

    Software
    Node.jsPython 3.11+Git
    Optional
    Alpaca Paper Trading Account (Free)Groq API Key (Free tier for natural language debate)

    Zero-config sandbox mode is included out of the box: runs immediately with simulated market feeds without requiring any broker accounts or API keys.

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Network Access
    Environment Variables

    File Scopes

    killjoy/**
    reports/**
    tests/**
    Listed5 days ago

    What's inside

    Frequently Asked Questions

    KILLJOY solves 'AI optimism' by using a dedicated Kill Agent to challenge every trade. Unlike standard bots that seek reasons to buy, this agent attempts to dismantle its own thesis, ensuring only the most robust trades pass its 9-stage adversarial cycle.

    The skill requires a Python environment and an Alpaca account for paper trading. It is optimized for high-speed inference through Groq, though it can be adapted for other LLM providers that support structured output.

    The 8 risk gates are hard-coded, deterministic rules—like Max Risk per Trade and Buying Power Floors—that act as absolute circuit breakers. These gates override any AI decision, preventing the agent from executing trades that violate your core financial safety parameters.

    You receive the full Python source code for the agent, the 9-stage adversarial pipeline, and the 8-gate risk management framework. It also includes the capital preservation reporting module to track how much loss was avoided by vetoed trades.

    Currently, execution is limited to Alpaca Paper Trading to ensure user safety while testing the adversarial logic. You can use the agent's synthesised Greeks and strategy reports to inform live trades manually, but direct live execution is disabled by default.

    No. KILLJOY is strictly engineered for paper trading. The codebase hard-enforces ALPACA_PAPER=true. If live production trading keys or live endpoints are supplied, execution is unconditionally rejected.

    Not at all. KILLJOY includes an automatic Zero-Config Sandbox Mode. Out of the box, it generates realistic 20-bar market feeds and full OCC options chains with strikes, expirations, and Greeks for SPY, NVDA, AAPL, and custom tickers. You can test and run complete cycles without any external credentials.

    Most trading agents suffer from confirmation bias and bullish hallucinations. KILLJOY pairs the strategy generator with an adversarial Kill Agent whose only objective is to find flaws in the trade. It tests for IV crush, volume divergence, theta decay, and liquidity traps. Only proposals that survive the kill gauntlet are allowed to proceed to the risk engine.

    Never. KILLJOY uses a hybrid architecture: while AI agents debate qualitative theses, the 8 Risk Gates are 100% deterministic code. Parameters like maximum $1,000 risk per trade, -$800 daily loss circuit breaker, and 5% buying power caps cannot be modified or overridden by LLM prompts.

    Yes. All risk limits and symbol universes are configured via simple environment variables in .env or passed directly to the CLI using --symbol <TICKER> and --strategy <STRATEGY>.

    New in Trading & Investing