Name: .agents
Author: kensonantony-prog

Imported from kensonantony-prog/QuanTexa-Financial-v2 (.agents/AGENTS.md). Install upstream with npx skills add kensonantony-prog/QuanTexa-Financial-v2 --skill .agents. Copyright stays with the author.

Financial Integrity Rules (Non-Negotiable)

Minimize look-ahead bias: No indicator, signal, or regime label at time t may use data after t.
Execution Timing: A signal generated from data through bar T executes at bar T+1. Never assume a fill at the same bar's close that produced the signal.
Transaction Costs: Transaction costs must reduce cash and portfolio value on every fill.
Position Sizing: Position size must control actual exposure, not just be a label.
Simulation Accuracy: The backtester must simulate money (cash, shares, equity), not just draw buy/sell markers.
Single Metric Implementation: ONE implementation of each metric in src/risk/metrics.py. The strategy, benchmark, robustness, and regime modules all call it. Never re-implement Sharpe, volatility, drawdown, or return elsewhere.
No Over-Optimization Bias: Never label the highest-return parameter set as "best" or "optimal". Robustness output shows the distribution of results and warns about over-optimization.
Visibility of Assumptions: All assumptions are visible in the UI and documented in research/assumptions.md.
Determinism: Same data + same parameters + same strategy = identical results (deterministic; no unseeded randomness).
Graceful Error Handling: Fail gracefully. Invalid input produces a clear message, never a crash.

QUANTEXA Design System Rules (Non-Negotiable)

Dark-First Command Center: Dark graphite (#05080D, #0A1018, #0E1621) with Electric Cyan (#19D3F3) accents.
No Playful SaaS Aesthetics: Avoid cartoon illustrations, bright rainbows, or soft consumer pill buttons everywhere.
Restrained Color Usage: Use semantic colors (#22C55E Success, #F59E0B Warning, #EF4444 Danger) strictly for actual risk and status communication.
Medium-High Information Density: Present dense financial metrics with tabular numbers (font-variant-numeric: tabular-nums).
Restrained Rounding: Card border-radius 10px, subtle 1px border (#1C2936), no heavy drop shadows.
Source of Truth: Follow tokens and principles in DESIGN.md.

{ "ocm": "1", "id": "kensonantony-prog-quantexa-financial-v2-agents", "kind": "skill", "name": ".agents", "description": "1. **Minimize look-ahead bias**: No indicator, signal, or regime label at time t may use data after t. 2. **Execution Timing**: A signal generated from data through bar T executes at bar T+1. Never assume a fill at the same bar's close that produced the signal. 3. **Transaction Costs**: Transaction costs must reduce cash and portfolio value on every fill. 4. **Position Sizing**: Position size must control actual exposure, not just be a label. 5. **Simulation Accuracy**: The backtester must simulate money (cash, shares, equity), not just draw buy/sell markers. 6. **Single Metric Implementation**: ONE implementation of each metric in `src/risk/metrics.py`. The strategy, benchmark, robustness, and regime modules all call it. Never re-implement Sharpe, volatility, drawdown, or return elsewhere. 7. **No Over-Optimization Bias**: Never label the highest-return parameter set as \"best\" or \"optimal\". Robustness output shows the distribution of results and warns about over-optimization. 8. **Visibility of Assumptions**", "publisher": "kensonantony-prog", "version": "1.0.0", "capabilities": { "domains": [ "finance" ], "tags": [ "skill-md", "github-agents" ], "languages": [ "en" ] }, "quality_prior": 0.6, "examples": [ "1. **Minimize look-ahead bias**: No indicator, signal, or regime label at time t may use data after t. 2. **Execution Timing**: A signal generated from data through bar T executes at bar T+1. Never assume a fill at the same bar's close that produced the signal. 3. **Transaction Costs**: Transaction costs must reduce cash and portfolio value on every fill. 4. **Position Sizing**: Position size must control actual exposure, not just be a label. 5. **Simulation Accuracy**: The backtester must simulate money (cash, shares, equity), not just draw buy/sell markers. 6. **Single Metric Implementation**: ONE implementation of each metric in `src/risk/metrics.py`. The strategy, benchmark, robustness, and regime modules all call it. Never re-implement Sharpe, volatility, drawdown, or return elsewhere. 7. **No Over-Optimization Bias**: Never label the highest-return parameter set as \"best\" or \"optimal\". Robustness output shows the distribution of results and warns about over-optimization. 8. **Visibility of Assumptions**" ], "primary": false, "metadata": { "source": { "provider": "github-agents", "repository": "https://github.com/kensonantony-prog/QuanTexa-Financial-v2", "path": ".agents/AGENTS.md", "ref": "1e0d82aa2e67628e72b76c29d5c4fb1a028717da", "url": "https://github.com/kensonantony-prog/QuanTexa-Financial-v2/blob/1e0d82aa2e67628e72b76c29d5c4fb1a028717da/.agents/AGENTS.md", "key": "kensonantony-prog/QuanTexa-Financial-v2/.agents/AGENTS.md" } }, "instructions": "# Financial Integrity Rules (Non-Negotiable)\n\n1. **Minimize look-ahead bias**: No indicator, signal, or regime label at time t may use data after t.\n2. **Execution Timing**: A signal generated from data through bar T executes at bar T+1. Never assume a fill at the same bar's close that produced the signal.\n3. **Transaction Costs**: Transaction costs must reduce cash and portfolio value on every fill.\n4. **Position Sizing**: Position size must control actual exposure, not just be a label.\n5. **Simulation Accuracy**: The backtester must simulate money (cash, shares, equity), not just draw buy/se", "cost": { "context_tokens": 538 } }

.agents

About

Financial Integrity Rules (Non-Negotiable)

QUANTEXA Design System Rules (Non-Negotiable)

Use it

Manifest

Reviews