content-consolidation

name: content-consolidation description: "Merges, deduplicates, and filters 5 interim files for enterprise relevance. Use when running Phase 1C of the newsletter pipeline. Takes 5 source-specific interim files from Phase 1B, consolidates into 30-50 enterprise-relevant discoveries. Keywords: consolidation, phase 1c, deduplication, enterprise filter, discoveries." metadata: category: domain phase: "1C"

Content Consolidation

Merge, deduplicate, and filter 5 Phase 1B interim files into a single discoveries document.

Quick Start

1. Read all 5 interim files from Phase 1B
2. Aggregate items into a working list with source tracking
3. Deduplicate: same feature from multiple sources becomes one item
4. Categorize into 5 taxonomy categories
5. Filter for enterprise relevance (score 5+)
6. Write output to workspace/newsletter_phase1a_discoveries_YYYY-MM-DD_to_YYYY-MM-DD.md

Inputs

All 5 Phase 1B interim files (required):

• workspace/newsletter_phase1b_interim_github_*.md
• workspace/newsletter_phase1b_interim_vscode_*.md
• workspace/newsletter_phase1b_interim_visualstudio_*.md
• workspace/newsletter_phase1b_interim_jetbrains_*.md
• workspace/newsletter_phase1b_interim_xcode_*.md

Output

• File: workspace/newsletter_phase1a_discoveries_YYYY-MM-DD_to_YYYY-MM-DD.md
• Target: 20-40 items with balanced category distribution

Legacy naming note: The output filename uses phase1a_discoveries despite being produced by Phase 1C. This is a historical convention preserved for compatibility with benchmark data, archived intermediates, and downstream prompt references. All downstream consumers (content-curation, prompts, agent) reference this exact name pattern.

Core Workflow

Key finding (see reference/source-intelligence/meta-analysis.md): Cross-cycle analysis shows the pipeline does NOT cut items. 100% discovery survival in Aug+Jun+May 2025. The real editorial value is in ADDITIONS (Azure, devblogs, resources), BUNDLING (models, parity, governance), and EXPANSION (flagships). This phase should ENRICH and CONSOLIDATE, not just filter.

Step 1: Aggregate

Read all 5 interim files. For each item, record: source file, date, title, URLs, status (GA/PREVIEW), relevance score, IDE support.

Step 1.5: Enrich (Gap-Filling)

Check for known under-discovery gaps (L29, L30, L31). These sources are consistently missing from Phase 1B output but always appear in published newsletters:

• Visual Studio devblogs: Search GitHub Changelog for entries containing "Visual Studio". Check devblogs.microsoft.com/visualstudio/ for monthly Copilot updates.
• Azure ecosystem: Check devblogs.microsoft.com/devops/ for Azure DevOps migration content, Azure MCP Server, Copilot for Azure updates.
• GitHub news-insights: Check github.blog/news-insights/company-news/ for strategic CPO/CEO announcements not in the changelog feed.
• Enablement resources: Check resources.github.com for new rollout playbooks, training content, Copilot Fridays additions.

Step 2: Deduplicate

See consolidation-rules.md for full dedup logic.

Key rules:

• Same feature from multiple sources: merge into single item, keep earliest date, combine all URLs, note IDE support breadth
• Same feature with different dates: single item with rollout timeline
• Similar but distinct features (e.g., "Auto model for VS Code" vs "Auto model for JetBrains"): keep separate but note relationship

Step 3: Categorize

Apply the 5-category taxonomy with target counts:

Step 4: Enterprise Relevance Filter

Apply relevance scoring (1-10 scale):

• 9-10: Must include (GA enterprise features, major security)
• 7-8: Strong candidate (significant updates)
• 5-6: Include if space permits
• 1-4: Exclude

Always exclude: Copilot Free/Individual/Pro/Pro+ features, consumer-only features, minor bug fixes.

Step 5: Label and Validate

• Apply consistent status labels: (GA), (PREVIEW), (EXPERIMENTAL), (DEPRECATED), (RETIRING)
• Add IDE support matrix using checkmarks
• Cross-reference: model releases should appear in Changelog + IDE notes; feature GA should appear in Blog + Changelog
• Verify all items are within DATE_RANGE and Reference Year

Step 5.5: Coverage Audit

Check that ALL standard newsletter categories have been assessed. If a category that appeared in recent newsletters has ZERO items, explicitly note: [Category]: No updates found in DATE_RANGE (verified: checked [sources]). This prevents silent omission of recurring sections.

Standard categories to audit: Security Updates, Code Quality, Platform Updates, Enterprise Admin, IDE Updates, Copilot Features, Deprecations.

Step 6: Write Output

Write the discoveries file with:

• Coverage summary table (source, item count, date range)
• Items organized by category
• Each item with full metadata
• Validation results checklist

Reference

• Consolidation Rules - Dedup logic, scoring, taxonomy
• Editorial Intelligence - Theme detection, selection weights, expansion/compression rules
• Benchmark Example - Known-good Dec 2025 discoveries

Done When

• Discoveries file exists at workspace/newsletter_phase1a_discoveries_*.md
• Total items: 20-40
• Category distribution within target ranges
• Zero duplicate items across sources
• No consumer-plan items included
• All items have source URLs (markdown links)
• All items within DATE_RANGE
• IDE coverage from all 5 sources represented