feat: Automate nightly news article generation with MCP integration

[Copilot]

Manual article creation for 14 languages consumed 2–4 hours daily and was error-prone. This adds a fully automated nightly pipeline: fetch → threshold → generate → validate → PR.

New files

• scripts/generate-daily-news.js — Orchestrator that queries riksdag-regering-mcp for documents published since yesterday, groups by type (bet/prop/mot), skips types below the minimum document threshold (default ≥5), and delegates HTML generation to the existing generate-news-enhanced.js engine. Week Ahead is always generated.

  # Typical nightly invocation (all 14 languages, threshold=5)
  node scripts/generate-daily-news.js --languages=all --threshold=5
  
  # Override date window, restrict types, dry-run
  node scripts/generate-daily-news.js --date=2026-02-18 --types=committee-reports --dry-run

• .github/workflows/nightly-news-generation.yml — Scheduled at 02:00 CET (0 1 * * *). All actions SHA-pinned. Supports workflow_dispatch with inputs for date, threshold, languages, types, and dry_run. Runs HTMLHint on generated files, updates indexes/sitemap, and opens a labeled PR via peter-evans/create-pull-request@v8.1.0.

• ARTICLE_ENHANCEMENT_GUIDE.md — Definitive 892-line reference consolidating patterns from Issues Enhance Committee Reports Articles with Full Data Analysis and Commentary (2026-02-10 to 2026-02-14) #306–334 (176 articles enhanced). Covers: all 32 MCP tool call examples, 150–400 word quality standards, The Economist style guidelines, 14-language translation order and RTL requirements, Schema.org synchronization requirements, index/sitemap update process, and 8 documented pitfalls (cold-start handling, professional translation safety guards, 1 MB patch-size limit, etc.).

package.json

Added generate-daily-news npm script entry pointing to the new orchestrator.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Automate Nightly News Article Generation with MCP Integration</issue_title>
<issue_description># 📋 Issue Type
Automation / Process Improvement

🎯 Objective

Create automated nightly news generation script that fetches latest Riksdag documents via MCP and generates enhanced articles with proper analysis automatically, eliminating manual article creation.

📊 Current State

Manual Process: Articles are created manually or semi-manually:

• Identify new documents (propositions, motions, committee reports)
• Fetch data via MCP
• Generate analysis
• Create 14 language versions
• Update indexes and sitemap

Problem: Time-consuming, error-prone, inconsistent quality

🚀 Desired State

Automated Workflow:

// Runs nightly at 02:00 CET via GitHub Actions
1. Fetch new documents from riksdag-regering MCP (last 24 hours)
2. Group by type (committee reports, propositions, motions)
3. Generate articles if threshold met (≥5 documents per type)
4. Apply content enhancement (150-400 word analysis)
5. Generate content-based titles
6. Create all 14 language versions
7. Update news indexes
8. Update sitemap.xml
9. Create PR with changes
10. Notify maintainers

🔧 Implementation Approach

Step 1: Create Generation Script

File: scripts/generate-daily-news.js

// Workflow:
// 1. Query MCP for new documents (last 24 hours)
riksdagRegering.searchDokument({
  rm: "2025/26",
  dateFrom: "YYYY-MM-DD" // yesterday
})

// 2. Group by document type
const committeeReports = docs.filter(d => d.doktyp === 'bet')
const propositions = docs.filter(d => d.doktyp === 'prop')
const motions = docs.filter(d => d.doktyp === 'mot')

// 3. Generate article if threshold met (≥5 docs)
if (committeeReports.length >= 5) {
  generateArticle({
    type: 'committee-reports',
    date: today,
    documents: committeeReports
  })
}

// 4. Apply enhancement template
// 5. Translate to 14 languages
// 6. Update indexes and sitemap
// 7. Commit and push

Step 2: GitHub Actions Workflow

File: .github/workflows/nightly-news-generation.yml

name: Nightly News Generation

on:
  schedule:
    - cron: '0 2 * * *'  # 02:00 CET daily
  workflow_dispatch:  # Allow manual trigger

jobs:
  generate-news:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with:
          node-version: '24'
      
      - name: Install dependencies
        run: npm install
      
      - name: Generate daily news
        run: node scripts/generate-daily-news.js
        env:
          MCP_AUTH_TOKEN: ${{ secrets.MCP_AUTH_TOKEN }}
      
      - name: Create PR
        uses: peter-evans/create-pull-request@v6
        with:
          title: "📰 Daily news: ${{ env.DATE }}"
          body: "Automated daily news generation"
          branch: "auto/daily-news-${{ env.DATE }}"

Step 3: Quality Controls

• Minimum document threshold (≥5 per type)
• MCP data validation
• HTML validation before commit
• Link checking
• Fallback to manual if errors

✅ Acceptance Criteria

• Script scripts/generate-daily-news.js created
• GitHub Actions workflow .github/workflows/nightly-news-generation.yml
• MCP integration working (fetch, analyze, enhance)
• 14-language generation automated
• Index/sitemap updates automated
• PR creation automated
• Error handling with notifications
• Manual trigger option available
• Documentation in ARTICLE_ENHANCEMENT_GUIDE.md

📚 References

• Existing scripts: scripts/generate-news-enhanced.js (partial implementation)
• MCP config: .github/copilot-mcp.json
• Enhancement patterns: Issues Enhance Committee Reports Articles with Full Data Analysis and Commentary (2026-02-10 to 2026-02-14) #306-311
• Translation workflow: scripts/translate-committee-reports.py

🤖 Recommended Agent

devops-engineer or data-pipeline-specialist

📝 Notes

High value: Eliminates 2-4 hours of manual work daily. Ensures consistent quality and timely publication. Reduces human error.

Dependencies: Should implement after Issue #338 (documentation) to follow proven patterns.

Timeline: 3-5 days</issue_description>

<agent_instructions>Create comprehensive ARTICLE_ENHANCEMENT_GUIDE.md documenting the complete workflow from Issues #306-334. Include: step-by-step process, all 32 MCP tool examples, content quality standards (150-400 words), translation workflow for 14 languages, validation steps, best practices, and common pitfalls. Reference successful PRs #307, #312, #313, #314, #326, #333, #334. Make it a definitive guide for future article generation.</agent_instructions>

Comments on the Issue (you are @copilot in this section)

• Fixes Automate Nightly News Article Generation with MCP Integration #338

---

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.