Files
gh-animalzinc-claude-plugin…/commands/batch-review.md
2025-11-29 17:55:01 +08:00

211 lines
5.9 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
---
description: Batch review multiple blog articles against editorial style guide
argument-hint: [brand-name] [directory-path] [--auto-apply]
---
# Batch Review: Process Multiple Articles
Review all articles in **$2** against the **$1** brand style guide.
## Step 1: Validate Inputs
Before proceeding:
- Verify style guide exists at `brands/$1/style-guide.md`
- If missing: Error and suggest running `/generate-style-guide $1 [source]`
- Verify directory exists at $2
- If missing: Error with clear message
- Check if $3 is `--auto-apply` (determines whether to apply corrections automatically)
## Step 2: Discover Articles
Scan the directory at $2 for markdown files:
- Use Glob to find all `*.md` files in $2
- Exclude files ending with `_edited.md` (already processed)
- Exclude README.md files
- Sort by filename
- Report number of articles found
**If no articles found:**
- Provide clear message and exit
**If articles found:**
- Display list of articles to be reviewed
- Ask user to confirm: "Found [X] articles. Proceed with batch review? (yes/no)"
## Step 3: Process Each Article
For each article in the list:
### Option A: Sequential Processing (Default - More Reliable)
1. Display: "Processing article [N] of [total]: [filename]"
2. Run the equivalent of `/style-check $1 [article-path]` on this article
3. Collect findings
4. If --auto-apply: Apply corrections and save to `edited/` folder
5. Move to next article
### Option B: Parallel Processing (Faster but Resource-Intensive)
- If user confirms they want parallel processing:
- Batch articles into groups of 3
- Process each group in parallel
- Note: This will spawn 24 agents per batch (8 agents × 3 articles)
## Step 4: Aggregate Portfolio Results
After all articles processed, create portfolio compliance report:
```markdown
# Batch Review Report: $1
**Brand:** $1
**Source Directory:** $2
**Review Date:** [current date]
**Articles Reviewed:** [number]
**Auto-Apply:** [yes/no]
---
## Portfolio Compliance Summary
**Overall Statistics:**
- Total articles reviewed: [X]
- Fully compliant (0 violations): [X] articles ([%])
- Minor issues (1-5 violations): [X] articles ([%])
- Moderate issues (6-15 violations): [X] articles ([%])
- Major issues (16+ violations): [X] articles ([%])
**Total Violations Across Portfolio:**
- Voice & Tone: [X] violations
- Grammar & Usage: [X] violations
- Punctuation: [X] violations
- Formatting: [X] violations
- Technical Standards: [X] violations
- Content Patterns: [X] violations
- Industry Terms: [X] violations
**Average violations per article:** [number]
---
## Most Common Violations
1. [Most frequent issue]: [X] occurrences across [Y] articles
2. [Second most frequent]: [X] occurrences across [Y] articles
3. [Third most frequent]: [X] occurrences across [Y] articles
4. [etc.]
---
## Article-by-Article Summary
| Article | Total Violations | Status | Priority |
|---------|-----------------|--------|----------|
| [filename-1] | [X] | ✅/⚠️/❌ | High/Med/Low |
| [filename-2] | [X] | ✅/⚠️/❌ | High/Med/Low |
| [etc.] | [X] | ✅/⚠️/❌ | High/Med/Low |
**Legend:**
- ✅ Fully compliant (0 violations)
- ⚠️ Minor issues (1-5 violations)
- ❌ Needs attention (6+ violations)
---
## Priority Recommendations
### High Priority Articles (Review First)
[List articles with major brand voice/structural issues]
### Medium Priority Articles
[List articles with consistency/formatting issues]
### Low Priority Articles
[List articles with minor polish needed]
---
## Common Patterns to Address
**Voice & Tone Issues:**
[List common voice/tone violations seen across multiple articles]
**Grammar & Usage Issues:**
[List common grammar violations]
**Formatting Issues:**
[List common formatting violations]
[etc.]
---
## Next Steps
1. Review high-priority articles first
2. Consider updating style guide if new patterns emerge
3. For articles with 0 violations: Consider them as exemplars
4. For articles with many violations: May indicate style guide misalignment
**Files Generated:**
- Batch report: docs/logs/[date]_batch-review_$1.md
- Individual reports: docs/logs/[date]_style-check_$1_[article].md (for each article)
- Edited articles: $2/../edited/ (if --auto-apply used)
```
## Step 5: Save Reports
Save the following:
- **Portfolio report**: `docs/logs/[YYYY-MM-DD]_batch-review_$1.md`
- **Individual reports**: One per article in `docs/logs/`
- **Edited articles**: If --auto-apply, save to `brands/$1/articles/edited/` or `$2/../edited/`
## Step 6: Update Brand README
If `brands/$1/README.md` exists:
- Add note about this batch review
- Update "Last Reviewed" date
- Add summary statistics
If doesn't exist:
- Create basic README with review statistics
## Usage Examples
```bash
# Review all articles in directory (report only)
/batch-review animalz brands/animalz/articles/original/
# Review and automatically apply corrections
/batch-review animalz brands/animalz/articles/original/ --auto-apply
# Review articles from any directory
/batch-review hubspot /Users/name/Documents/hubspot-drafts/
```
## Performance Estimates
**Sequential processing (default):**
- Time per article: ~5-10 minutes
- For 10 articles: ~50-100 minutes (1-2 hours)
- More reliable, less resource-intensive
**Parallel processing (if user requests):**
- Time for 3 articles (batch): ~10-15 minutes
- For 10 articles (4 batches): ~40-60 minutes
- Faster but more complex
## Error Handling
- **Style guide missing**: Provide helpful error with suggestion
- **Directory not found**: Check path and provide clear error
- **No articles found**: Inform user and suggest checking directory
- **Individual article failures**: Continue with remaining articles; note failures in report
- **Permission issues**: Report which files couldn't be accessed
## Quality Assurance
After batch processing:
- Verify all articles were processed
- Check that reports were generated
- If --auto-apply: Verify edited files were created
- Count total violations vs. articles to ensure reasonable results