scenario-analyzer

The saved excerpt is a snapshot from review. The external source remains the complete and most current version.

---
name: scenario-analyzer
description: |
  Skill that analyzes 18-month scenarios from a news headline.
  Runs the primary analysis with the scenario-analyst agent and obtains a
  second opinion with the strategy-reviewer agent.
  Generates a comprehensive English report covering 1st/2nd/3rd-order
  impacts, recommended stocks, and a critical review.
  Example: /scenario-analyzer "Fed raises rates by 50bp"
  Triggers: news analysis, scenario analysis, 18-month outlook,
  medium-to-long-term investment strategy
---

# Scenario Analyzer

## Overview

This skill analyzes medium-to-long-term (18-month) investment scenarios
starting from a news headline. It invokes two specialized agents in sequence
(`scenario-analyst` and `strategy-reviewer`) and integrates multi-angle
analysis with a critical review into a comprehensive report.

## When to Use This Skill

Use this skill when:

- You want to analyze the medium-to-long-term investment impact of a news headline
- You want to construct multiple 18-month scenarios
- You want sector/stock impacts organized into 1st/2nd/3rd-order effects
- You need a comprehensive analysis that includes a second opinion

**Examples:**
```
/scenario-analyzer "Fed raises interest rates by 50bp, signals more hikes ahead"
/scenario-analyzer "China announces new tariffs on US semiconductors"
/scenario-analyzer "OPEC+ agrees to cut oil production by 2 million barrels per day"
```

## Prerequisites

- **API Keys**: None (uses only WebSearch/WebFetch)
- **MCP Servers**: None
- **Dependencies**: The scenario-analyst and strategy-reviewer agents must be available via the Task tool

## Architecture

```
┌─────────────────────────────────────────────────────────────────────┐
│                    Skill (orchestrator)                              │
│                                                                      │
│  Phase 1: Preparation                                                │
│  ├─ Headline parsing                                                 │
│  ├─ Event type classification                                        │
│  └─ Reference loading                                                │
│                                                                      │
│  Phase 2: Agent invocation                                           │
│  ├─ scenario-analyst (primary analysis)                              │
│  └─ strategy-reviewer (second opinion)                               │
│                                                                      │
│  Phase 3: Integration & report generation                            │
│  └─ reports/scenario_analysis_<topic>_YYYYMMDD.md                   │
└─────────────────────────────────────────────────────────────────────┘
```

## Workflow

### Phase 1: Preparation

#### Step 1.1: Headline Parsing

Parse the headline provided by the user.

1. **Headline check**
   - Confirm a headline was passed as an argument
   - If not provided, ask the user for input

2. **Keyword extraction**
   - Key entities (company names, country names, institution names)
   - Numeric data (rates, prices, quantities)
   - Actions (raise, cut, announce, agree, etc.)

#### Step 1.2: Event Type Classification

Classify the headline into one of the following categories:

| Category | Examples |
|----------|----------|
| Monetary Policy | FOMC, ECB, BOJ, rate hike, rate cut, QE/QT |
| Geopolitics | War, sanctions, tariffs, trade friction |
| Regulation & Policy | Environmental regulation, financial regulation, antitrust |
| Technology | AI, EV, renewables, semiconductors |
| Commodities | Crude oil, gold, copper, agricultural products |
| Corporate & M&A | Acquisitions, bankruptcies, earnings, industry restructuring |

#### Step 1.3: Reference Loading

Based on the event type, load the relevant references:

```
Read references/headline_event_patterns.md
Read references/sector_sensitivity_matrix.md
Read references/scenario_playbooks.md
```

**Reference contents:**
- `headline_event_patterns.