review-content

name	review-content
description	Review content for tone, style, structure, glossary coverage, and linking completeness against the style guide.
argument-hint	<file-path>
user-invocable	true
allowed-tools	Read, Write, Edit, Glob, Grep, Bash, Task, TodoWrite

Review content for tone, style, structure, and linking completeness.

File: $ARGUMENTS

Pass the path to an article or glossary file (e.g., src/content/articles/en/my-article.md).

Instructions

The content-style and content-inventory skills auto-load. Read the target file to understand what you're reviewing.

Phase 1: Style Review via style-enforcer Agent

Read .content-ops/config.md to detect the content type from the file path (match against content_types.*.path).

Delegate the full style audit to the style-enforcer agent:

Use the style-enforcer agent to review the file at [path].
Content type: [detected content type]
Config:
  word_range: [content_types.<type>.word_range]
  guidelines: [content_types.<type>.guidelines list]
  reference_content: [reference_content list]
Perform a complete review.

The agent returns a structured report with score, must-fix, should-fix, and consider categories.

Phase 2: Glossary Coverage Audit

Skip this phase if the glossary config block is absent or glossary.enabled is false.

Read the glossary block from .content-ops/config.md to determine if glossary is enabled.

If enabled, scan the article body for domain-specific technical terms. For each:

1. Has glossary entry + is linked? — Good.
2. Has glossary entry + NOT linked? — Flag: "Term X is used but not linked to its glossary entry."
3. No glossary entry + is a technical term? — Flag: "Term X has no glossary entry. Consider creating one with /write-content glossary X."

Check that relatedGlossary in frontmatter includes terms linked in body. Note: relatedGlossary may also include terms relevant to the topic even without inline links — this is valid.

Phase 3: Linking Completeness Audit

1. Inline glossary links — Only on first mention? Flag duplicates.
2. Inline article links — Related articles linked in body where natural?
3. Frontmatter relatedArticles — Includes all genuinely related articles?
4. Frontmatter relatedGlossary — Includes all glossary terms linked in body?
5. Bidirectional check — For each reference, check if the referenced content links back. Flag missing back-links as recommendations (not mandatory — there are cases where a back-link may not make sense).

Phase 4: Glossary-Specific Checks (glossary files only)

Skip this phase if the glossary config block is absent or glossary.enabled is false.

Also skip if the file being reviewed is not the glossary content type (check file path against content_types.[glossary.content_type].path from config).

If applicable:

• Definition length — 1-2 sentences. Flag if longer.
• Example — One concrete sentence. Flag if missing or abstract.
• Body content — 1-2 dictionary-style sentences. Flag if longer.
• Term capitalization — Proper nouns capitalized, common terms lowercase.

Phase 5: Combined Report

Present the style-enforcer agent's report combined with your glossary and linking findings:

## Content Review: <title>

### Score: X/10 (from style-enforcer)

### Must Fix
- [Critical: style issues + broken links + missing glossary links for jargon]

### Should Fix
- [Style issues + missing back-links + duplicate links]

### Consider
- [Suggestions: glossary entries to create, additional linking, structural improvements]

### Looks Good
- [Strengths — always acknowledge what's working]

For each issue: exact location, what's wrong, concrete suggestion.

Phase 6: Fix (with permission)

After presenting the report, ask the user if they want you to fix any issues. If yes:

1. Apply fixes in severity order (must fix → should fix → consider)
2. For linking fixes, delegate to the content-linker agent
3. Commit: content: review fixes for "<title>"