Analyzed all 4 news generation workflows and identified critical issues blocking multi-language synchronization, quality validation, and workflow coordination. Created detailed implementation issues with measured current state, technical approaches, and test requirements.

Issues Created

#155 - news-realtime-monitor.md

• Multi-language index broken: news/index_fr.html only shows French articles
• Fix: ALL 14 index files must show ALL articles with language badges
• Add Economist-style quality framework (min score 0.75)
• Implement workflow coordination via news/metadata/workflow-state.json
• 20+ tests for breaking news criteria, cross-referencing, deduplication

#157 - news-evening-analysis.md

• Missing test coverage for 5 Editorial Pillars structure
• Add analytical depth validation (≥0.6 score required)
• Implement cross-workflow deduplication (>70% similarity threshold)
• Track quality metrics per article in news/metadata/quality-metrics.json
• 25+ tests for structure, depth, historical context

#159 - news-article-generator.md

• Monolithic code needs refactoring into 5 modular article types
• Validate cross-referencing patterns per type (3+ sources required)
• Automate Playwright visual validation with screenshot capture
• Create scripts/news-types/ modules: week-ahead, committee-reports, propositions, motions, breaking-news
• 60+ tests across all article types

#161 - news-generation.yml

• Critical bug: Timestamp commits on 51% of runs (should be <20%)
• Fix: Only commit timestamp when generated=0
• Add structured error logging to news/metadata/errors.json
• Integrate with agentic workflows (skip if recent activity <2h)
• Fix PR creation logic (only when articles generated)
• 20+ tests for timestamp logic, error handling, language expansion

Key Patterns Established

Multi-Language Index Architecture

// BEFORE (wrong): Filter by language
const frArticles = allArticles.filter(a => a.file.endsWith('-fr.html'));

// AFTER (correct): Show all articles with language badges
const allArticlesForIndex = allArticles.map(article => ({
  ...article,
  availableLanguages: detectAvailableLanguages(article.slug, allArticles)
}));

Quality Framework Requirements

• Score ≥0.75 (quantitative quality)
• "Why This Matters" section (100-150 words)
• 4+ party perspectives
• 3+ riksdag-regering sources with document IDs
• Historical context + international comparison

Workflow Coordination State

{
  "lastUpdate": "2026-02-14T10:00:00Z",
  "recentArticles": [/* last 6 hours */],
  "mcpQueryCache": {/* 2-hour TTL */}
}

Measured Current State

• Articles: 22 generated (Feb 10-14), 14 languages for evening analysis
• Workflows: 35 runs (80% success, 20% failure)
• Issues: 51% timestamp-only commits, 69% zero-article runs
• Gaps: Multi-language sync broken, no quality scoring, partial test coverage

Expected Outcomes

• 125+ comprehensive tests
• 14-language cross-discovery enabled
• Timestamp commits reduced 51% → <20%
• Quality score ≥0.75 for all articles
• Workflow success rate 80% → 95%

All issues assigned to Copilot with detailed custom instructions. No code changes in this PR - analysis and issue creation only.

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.