News Intelligence

News Intelligence Radar

Comprehensive situational awareness engine that gathers news across domains, identifies cross-domain connections, assesses impact, and routes actionable intelligence to your other skills.

When This Skill Triggers

• "What's happening?" / "Catch me up" / "Morning brief"
• "News about gold / forex / AI / tech"
• "What's affecting [instrument/market]?"
• "How does [event X] impact [domain Y]?"
• "Scan news for [topic]"
• "What should I be watching today?"
• Any question about current events that benefits from multi-source analysis
• References to market-moving events, geopolitical situations, or industry developments

Architecture

User Query
    │
    ├─[1] CLASSIFY REQUEST → What type of scan?
    │     ├── Morning Brief (all priority domains)
    │     ├── Topic Scan (specific subject)
    │     ├── Event Analysis (single event, deep)
    │     └── Domain Scan (one domain, thorough)
    │
    ├─[2] GENERATE SEARCH PLAN → QueryGenerator
    │     └── 4-12 targeted web_search queries
    │
    ├─[3] EXECUTE SEARCHES → web_search + web_fetch
    │     └── Collect titles, summaries, sources, URLs
    │
    ├─[4] PROCESS RESULTS
    │     ├── DomainClassifier → Tag each item by domain
    │     ├── EntityExtractor → Pull currencies, orgs, numbers
    │     ├── ImpactAssessor → Rate high/medium/low
    │     └── RelationshipMapper → Find cross-domain chains
    │
    ├─[5] ROUTE TO SKILLS → SkillRouter
    │     └── Recommend which skills to activate next
    │
    └─[6] FORMAT OUTPUT → IntelligenceBrief
          ├── Priority Alerts
          ├── Domain Sections
          ├── Cross-Domain Connections
          ├── Skill Route Suggestions
          └── Narrative (Claude-generated synthesis)

How to Execute This Skill

Step 0: Load the Engine

import sys
sys.path.insert(0, '/path/to/news-intelligence/scripts')
from news_engine import (
    QueryGenerator, DomainClassifier, EntityExtractor,
    ImpactAssessor, SkillRouter, IntelligenceBrief,
    generate_search_plan, get_upcoming_events, DOMAINS
)
from relationship_mapper import RelationshipMapper, build_narrative_prompt

Step 1: Determine Scan Type

Based on what the user asked:

Step 2: Generate & Execute Search Plan

plan = generate_search_plan(mode="morning_brief")
# plan["queries"] → list of {"query": str, "domain": str}
# plan["upcoming_events"] → scheduled economic events today

# Then execute each query using web_search tool
# Collect results into news_items list

CRITICAL — Search Execution Pattern:

For each query in the plan, Claude should:

1. Call web_search with the query
2. Read the top 3-5 results
3. For high-impact items, call web_fetch to get the full article
4. Build a news_item dict for each meaningful result:

news_item = {
    "title": "...",
    "source": "Reuters",
    "summary": "2-3 sentence summary in your own words",
    "url": "https://...",
    "domain": "",  # Will be classified
    "entities": {},  # Will be extracted
    "impact_level": "",  # Will be assessed
    "timestamp": "2025-01-01",
}

Step 3: Process Results

After collecting all news items:

for item in news_items:
    # Classify domain
    domains = DomainClassifier.classify(item["summary"], item["title"])
    if domains:
        item["domain"] = domains[0]["domain"]
        item["classified"] = domains

    # Extract entities
    item["entities"] = EntityExtractor.extract(f"{item['title']} {item['summary']}")

    # Assess impact
    item["impact_level"] = ImpactAssessor.assess(item["title"], item["summary"])

Step 4: Find Connections & Route

# Cross-domain connections
connections = RelationshipMapper.find_connections(news_items)
shared = RelationshipMapper.find_shared_entities(news_items)

# Skill routing
all_domains = []
for item in news_items:
    all_domains.extend(item.get("classified", []))
skill_routes = SkillRouter.route(all_domains)

Step 5: Format the Brief

brief = IntelligenceBrief.format_brief(
    news_items=news_items,
    skill_routes=skill_routes,
    brief_type="morning"
)

# Add connections report
connections_report = RelationshipMapper.format_connections_report(connections, shared)

Step 6: Write the Narrative (This is YOU, Claude)

After formatting the structured brief, write a 3-4 paragraph narrative that:

• Tells the story connecting the dots across domains
• Highlights the single most important development
• Explains cascading effects (e.g., "Fed dovish → USD weak → Gold up → trading opportunity")
• Suggests 2-3 concrete actions linked to specific skills

Use the build_narrative_prompt() function from relationship_mapper.py as a starting guide, but you should write the narrative directly since YOU have all the context.

Covered Domains

Economic Calendar Integration

The engine includes awareness of recurring high-impact events:

• US NFP (first Friday), CPI, FOMC (mid-month)
• ECB/BOJ rate decisions
• Weekly jobless claims (Thursday), oil inventory (Wednesday)

Call get_upcoming_events() to check if any are scheduled today. Always mention relevant upcoming events in the brief — they affect how to interpret current news.

Output Formats

Morning Brief (Default)

Full structured brief with priority alerts, domain sections, connections, and skill routes.

Quick Scan

Single-topic focused scan. Use IntelligenceBrief.format_quick_scan(topic, findings).

Event Analysis

Deep dive into a single event with multi-angle coverage and impact assessment.

Mermaid Relationship Graph

Visual graph of cross-domain connections. Use RelationshipMapper.generate_mermaid_graph().

Practical Examples

"What's happening with gold today?"

1. Search: gold price today, XAUUSD analysis, gold market drivers
2. Also search: Fed news (macro driver), geopolitics (safe haven driver)
3. Classify, connect (geopolitics → gold, rates → gold)
4. Route to: automated-strategy-builder, order-block-mapper
5. Output: Quick scan + trading implications

"Morning brief"

1. Search across: forex, gold, AI, geopolitics (12-16 queries)
2. Process all results through the full pipeline
3. Output: Full intelligence brief with narrative

"How does the new AI regulation affect my job search?"

1. Search: AI regulation news, AI job market impact, AI engineer demand
2. Classify: ai_tech + freelance_career
3. Connect: regulation → hiring patterns → freelance demand
4. Output: Event analysis with career-specific implications

Script Locations

news-intelligence/
├── SKILL.md                              ← You are here
├── scripts/
│   ├── news_engine.py                    ← Core engine (queries, classify, assess, route, format)
│   └── relationship_mapper.py            ← Cross-domain connections & causal chains
└── references/
    └── domain_deep_dives.md              ← Extended domain analysis templates

Tips for Best Results

1. Scale searches to query complexity: Morning brief = 12-16 searches. Quick topic scan = 4-6.
2. Always check the economic calendar via get_upcoming_events() — scheduled events change everything.
3. Use web_fetch for high-impact items — search snippets are often too brief for good analysis.
4. Write the narrative yourself — don't just dump the structured data. The user wants your synthesis.
5. Route to skills proactively — if gold is moving, suggest running the strategy builder without being asked.
6. Cross-domain connections are the value add — anyone can search news. The skill is connecting dots.