notque/vexjoy-agent / skills/process/workflow
workflow
VerifiedStructured work: multi-phase tasks, feature builds, planning, objective loops, hill climbing.
references/article-evaluation-pipeline.md
View on GitHub---
description: "Wabi-sabi-aware 4-phase article evaluation for voice authenticity."
user-invocable: false
allowed-tools:
- Read
- Write
- Bash
- Grep
- Glob
- Edit
- Task
- WebFetch
context: fork
routing:
triggers:
- evaluate article
- check article voice
- is this authentic
- review my article
- voice evaluation
- article evaluation
- check my voice
pairs_with:
- writing
complexity: medium
category: content
---
# Article Evaluation Pipeline Skill
## Overview
This skill evaluates articles for voice authenticity through a deterministic 4-phase pipeline: **Fetch**, **Validate**, **Analyze**, **Report**. It combines voice pattern validation via `voice-validator.py` with wabi-sabi-aware analysis to distinguish authentic imperfections from actual violations. The evaluation produces a structured report with verdict (AUTHENTIC / NEEDS WORK / FAILED) and line-specific recommendations.
**Default behaviors (always enabled)**:
- All 4 phases execute in sequence with explicit gates between phases
- Voice profile auto-detection from source context or manual `--voice` override
- Banned pattern zero-tolerance checking alongside voice validation
- Wabi-sabi analysis distinguishing natural imperfections from AI tells
- Line number attribution for all findings
- Evaluation report saved to file artifact
**Optional behaviors (off unless triggered)**:
- Quick mode: Skip wabi-sabi analysis, report scores only (`--quick`)
- Fix suggestions: Generate revision items for NEEDS WORK/FAILED verdicts (`--suggest-fixes`)
- Force voice: Override auto-detection with explicit profile (`--voice {name}`)
**Scope boundaries**: This skill evaluates existing articles only. Do not use for writing (use writing), editing (use the private de-AI editor skill), or creating voice profiles (use writing).
---
## Instructions
### Phase 1: FETCH
**Goal**: Obtain article content and identify the target voice profile.
**Step 1: Identify source**
Determine whether input is a URL or local file path.
**Step 2: Fetch content**
- For URLs: Use WebFetch or curl to retrieve content, extract article body as markdown
- For local files: Read the file directly using Read tool
**Step 3: Save artifact**
Store content to `/tmp/article-evaluation-[timestamp].md` for use in subsequent phases. This persistent artifact is non-negotiable — all downstream validators reference this file.
**Step 4: Detect voice**
Identify voice profile from source context:
- Known domain/path mapping → fetch corresponding voice profile
- Unknown source → require explicit `--voice {name}` flag
Do not guess voice profiles. Wrong profile produces meaningless scores and invalidates downstream analysis.
**Gate**: Article content exists in temp file AND voice profile identified. Proceed only when gate passes.
### Phase 2: VALIDATE
**Goal**: Run deterministic validation against voice profile and banned patterns using `skills/content/voice-validator/scripts/voice-validator.py`.
**Step 1: Voice pattern validation**
```bash
python3 $HOME/vexjoy-agent/skills/content/voice-validator/scripts/voice-validator.py validate \
--content /tmp/article-evaluation.md \
--voice [voice-name] \
--format json
```
Pass criteria: Score >= 60, zero hard errors.
**Step 2: Banned pattern check**
```bash
python3 $HOME/vexjoy-agent/skills/content/voice-validator/scripts/voice-validator.py check-banned \
--content /tmp/article-evaluation.md
```
Pass criteria: Score = 100 (no banned patterns detected).
Em-dashes are absolutely prohibited and must always be flagged as errors. This is a zero-tolerance policy.
**Step 3: Record results**
Capture both validation scores, all errors with line numbers, and all warnings. Trust the script output over subjective assessment — deterministic validation is non-negotiable.
**Gate**: Both validation runs complete with output captured and line numbers recorded. Proceed only when gate passes.
### Phase 3: ANALYZE (Wabi-Sabi)
**Goal**: Classify imperfections as authentic markers or actual violations. This phase distinguishes whether deviations from "perfect" writing are intentional stylistic features or genuine problems.
**Step 1: Scan for imperfections**
Review content for all deviations from grammatical perfection: typos, run-ons, fragments, self-corrections, trailing thoughts, casual contractions, natural rhythm breaks.
**Step 2: Classify each finding**
For each imperfection found, classify as one of:
- **WABI-SABI** (KEEP): Intentional imperfection matching the writer's authentic patterns. Natural-sounding deviations that enhance authenticity.
- **ERROR** (FIX): Actual voice violation, banned pattern, or content that contradicts the authentic voice.
- **WARNING** (REVIEW): Minor rhythm issue or pattern inconsistency worth documenting but not blocking.
Use `references/wabi-sabi-classification.md` for the full decision tree. Do not flag typos, run-ons, or fragments as errors automatically — evaluate them against the writer's authentic voice patterns first.
**Step 3: Check for suspicious perfection**
Zero wabi-sabi markers is itself a red flag. If no imperfections found, note as suspicious — authentic writing always contains natural imperfections. Over-polished content suggests synthetic generation.
**Gate**: All imperfections classified with line numbers and rationale. Document the wabi-sabi verdict (markers present / absent). Proceed only when gate passes.
### Phase 4: REPORT
**Goal**: Generate comprehensive evaluation report with verdict and recommendations.
**Step 1: Compile findings**
Aggregate validation scores, wabi-sabi markers, errors, and warnings into the structured report format defined in `references/report-template.md`. All findings must include line numbers and context.
**Step 2: Determine verdict**
| Verdict | Conditions |
|---------|------------|
| AUTHENTIC | Voice >= 60, banned = 100, wabi-sabi markers present |
| NEEDS WORK | Voice >= 60, banned < 100 (minor violations) |
| FAILED | Voice < 60, or major banned pattern violations |
A passing score is not about over-polishing (95+). Expect authentic articles to score 70-90 with visible wabi-sabi markers. Scores below 60 or banned pattern violations require remediation.
**Step 3: Write recommendations**
- For NEEDS WORK or FAILED verdicts: List specific items to fix with line numbers and rationale
- For AUTHENTIC verdicts: Document what makes it work — which wabi-sabi markers contribute to authenticity
**Step 4: Output report**
Display report to user and save to file. The report must include verdict, both scores, wabi-sabi analysis, and specific recommendations.
**Gate**: Complete report generated with verdict, scores, wabi-sabi analysis, and recommendations. Evaluation complete.
---
## Examples
### Example 1: URL Evaluation
User says: "Evaluate this article https://example.com/posts/my-article/"
Actions:
1. Fetch article content, save to temp file, detect voice from context (FETCH)
2. Run voice-validator.py validate + check-banned (VALIDATE)
3. Classify imperfections as wabi-sabi or violations (ANALYZE)
4. Generate report with verdict (REPORT)
Result: Comprehensive evaluation with AUTHENTIC/NEEDS WORK/FAILED verdict and line-specific findings
### Example 2: Local File Quick Check
User says: "Quick check on ~/myblog/content/posts/draft.md"
Actions:
1. Read local file, save to temp, detect voice from context (FETCH)
2. Run both validators (VALIDATE)
3. Skip wabi-sabi analysis (quick mode enabled) (ANALYZE skipped)
4. Generate abbreviated report with scores only (REPORT)
Result: Fast pass/fail with scores, no wabi-sabi breakdown
---
## Error Handling
### Error: "Voice validator script not found"
Cause: `skills/content/voice-validator/scripts/voice-validator.py` not at expected path or not executable
Solution:
1. Verify path: `ls $HOME/vexjoy-agent/skills/content/voice-validator/scripts/voice-validator.py`
2. Check permissions: `chmod +x` if needed
3. If missing, cannot proceed — deterministic validation via the script is a non-negotiable requirement
### Error: "Cannot determine voice profile"
Cause: Source does not match any known site mapping and no `--voice` flag provided
Solution:
1. Ask user which voice to validate against
2. Use `--voice {name}` explicit flag
3. Use the explicit `--voice {name}` flag so the evaluation uses the right profile; guessing here produces meaningless scores and invalidates the entire evaluation
### Error: "Article content empty or too short"
Cause: WebFetch failed, URL is paywalled/auth-walled, or file path incorrect
Solution:
1. Verify URL is accessible (check for paywalls, auth walls, redirects)
2. Try alternative fetch method (curl vs WebFetch)
3. Ask user to provide content directly if URL remains inaccessible
---
## References
- `${CLAUDE_SKILL_DIR}/references/report-template.md`: Full report format with verdict criteria and examples
- `${CLAUDE_SKILL_DIR}/references/wabi-sabi-classification.md`: Complete marker tables and classification decision tree
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies. | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |
| meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要". | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |