Coalesce Articles

Merge multiple related articles into a single unified piece, archiving the originals to preserve their URLs for external links.

When to Use

When explicitly invoked: /coalesce
When multiple articles have significant overlap
When a topic has fragmented into too many small pieces
After identifying redundancy during review

Arguments

Provide as space-separated arguments or answer when prompted:

Source articles (comma-separated paths): obsidian/concepts/article-1.md,obsidian/concepts/article-2.md
Target location: obsidian/topics/new-unified-article.md

Example: /coalesce obsidian/concepts/a.md,obsidian/concepts/b.md obsidian/topics/combined.md

Instructions

1. Validate Inputs

Verify all source files exist and target location is valid:

Source files must exist in obsidian/
Target must be in a valid sync directory (topics/, concepts/, voids/, etc.)
Target should not already exist (or ask for confirmation to merge into existing)
Check for slug collisions before creating the target file:
```
uv run python scripts/check_slug.py <target-slug> <target-section>
```
If the check reports a collision, choose a different filename.

If arguments are not provided, check section_caps in obsidian/workflow/evolution-state.yaml and count .md files (excluding index files) in obsidian/topics/, obsidian/concepts/, and obsidian/voids/. Prioritize searching for merge candidates in the section closest to its cap (by percentage filled). This ensures coalesce creates room where it's most needed. If no good candidates exist in the most-pressured section, try the next, then search globally. Choose the most promising two or more candidates for merging. If there are no good candidates, abandon the attempt and do not merge anything.

2. Analyze Source Content

Read all source articles and identify:

Common themes and overlapping content
Unique contributions from each source
Conflicting claims or framings
Cross-references between sources

Also search for external references to these articles:

grep -r "[[source-slug" obsidian/

If you selected aricles automaticall but find that the articles ar well-differntiated and should not be merged, search again. If after five iterations there are still no strong candiates, abaon the attempt and do not merge anything.

3. Plan the Merge

Create a mental outline:

What is the unified article's core claim?
How do the sources' unique contributions organize into sections?
Which redundant content to eliminate?
How to reconcile any conflicts?

4. Generate Unified Article

Create the new article at the target location.

Frontmatter:

---
title: "[Unified Title]"
created: [earliest source created date]
modified: [today's date]
human_modified: [latest human_modified from sources, or null]
ai_modified: [current timestamp]
draft: false
topics: [merge from sources, deduplicate]
concepts: [merge from sources, deduplicate]
related_articles: [merge from sources, remove self-references, deduplicate]
ai_contribution: [calculate: if all sources were AI, use 100; if mixed, estimate]
author: [preserve if any human-authored sources]
ai_system: [current model]
ai_generated_date: [today]
last_curated: null
coalesced_from:
  - "/[section]/[source-1-slug]/"
  - "/[section]/[source-2-slug]/"
---

Content guidelines:

Follow writing-style.md
Front-load the key claim
Self-contained article
Include "Relation to Site Perspective" section
Include "Further Reading" section
Preserve valuable unique insights from each source
Eliminate redundancy

5. Archive Source Articles

For each source article:

Calculate archive path: archive/[original-section]/[slug].md
- Example: obsidian/concepts/old-article.md → archive/concepts/old-article.md
Move file to archive directory (preserving content)
Update frontmatter with archive metadata:

archived: true
archived_date: [current ISO timestamp]
superseded_by: "/[target-section]/[target-slug]/"
archive_reason: "Coalesced into [new article title]"
original_path: "/[original-section]/[original-slug]/"

The original content remains intact below the frontmatter. The archive notice will be displayed by Hugo templates.

6. Check for References

Search for wikilinks to the archived articles in active content:

grep -r "\[\[source-slug" obsidian/

For each reference found:

Note the file and context
These may need updating to point to the new unified article
Create a follow-up task if references found

7. Do Not Mark Todo Complete

Leave the originating todo task untouched. The orchestrator (cycle_post.py) marks it complete after this skill returns.

If references were found to archived articles, create a follow-up task:

### P2: Update references to coalesced articles
- **Type**: cross-review
- **Notes**: Coalesce created [new-article]. The following files reference archived articles and may need review: [list files]
- **Source**: coalesce
- **Generated**: [today]

8. Log to Changelog

Prepend to obsidian/workflow/changelog.md (add immediately after frontmatter, before existing entries):

## [current time from prompt] - coalesce
- **Status**: Success
- **Sources**: [list of source files as wikilinks]
- **Target**: [[new article path without .md extension]]
- **Archived**: [list of archive paths as wikilinks]
- **References to review**: [count or "none"]

9. Do Not Commit

Leave file changes uncommitted on disk. The orchestrator (cycle_post.py for /loop runs, agent-commit for legacy evolve_loop.py runs) creates the git commit after this skill returns. For manual invocations, run /agent-commit afterward or commit by hand.

Important

ALWAYS preserve source content in archive (never delete)
ALWAYS update frontmatter with archival metadata
The original URLs will continue to work via the archive
Check for external references before archiving

coalesce

SKILL.md.claude/skills/coalesce