ai-paper-review

Files (1)

• SKILL.md

SKILL.md (first lines)

Review a student paper and insert Word comments with substantive feedback tied to the assignment rubric.

Target file: $ARGUMENTS

## Supported Assignments

### ai-leading-learning (default)
**AI: Leading and Learning** — 8-10 page APA 7th edition paper with five pillars.

#### Five Pillars to Evaluate

1. **Define & Contextualize**: Does the paper define AI in the K-12 landscape? Does it go beyond "what it is" to explain "why it matters" for school improvement and instructional leadership?

2. **Research & Policy Review**: Are there at least five scholarly or credible sources synthesized (not just summarized)? Does it address AI's impact on learning outcomes, teacher workload, and data privacy?

3. **Instructional Leadership Analysis**: Are AI implementation strategies explicitly connected to TILS Standards A-D? Are specific standard indicators cited, not just general references?

4. **Risks & Ethical Considerations**: Does the paper critically examine algorithmic bias, the digital divide in access, and student data protection under FERPA?

5. **Evidence-Based Recommendations**: Are there 3-5 actionable strategies? Are they aligned with a school mission? Do they promote equitable student success? Are they grounded in the research reviewed earlier in the paper?

#### Criteria for Success (comment on these)

- **APA Formatting**: Correct citations, headings (Level 1 and 2), running head, references page, 12pt Times New Roman or similar, double spacing, 1-inch margins
- **Critical Synthesis**: Is the student comparing and contrasting perspectives, or just summarizing sources one by one?
- **Standard Alignment**: Are connections between AI strategies and specific TILS/PSEL indicators clear and explicit?
- **Actionability**: Are recommendations realistic and grounded in the research reviewed, or generic/aspirational?
- **Page Length**: Is the paper within the 8-10 page requirement (not counting title page and references)?

## Comment Categories

When reviewing, organize feedback into these categories and prefix each comment accordingly:

- **[Strength]** — Something done well; reinforce good practice
- **[Content]** — Missing content, shallow treatment, or factual issues
- **[Synthesis]** — Where the student is summarizing instead of synthesizing; suggest how to compare/contrast
- **[TILS]** — Missing or weak connections to specific TILS Standards A-D; note which standard applies
- **[APA]** — Formatting, citation, or reference errors
- **[Evidence]** — Claims made without supporting citations, or recommendations not grounded in the reviewed research
- **[Ethics]** — Gaps in the risks/ethical analysis (bias, digital divide, FERPA)
- **[Clarity]** — Writing quality issues: awkward phrasing, vague language, unclear arguments
- **[Typo]** — Spelling or grammar errors

## TILS Standards Reference (for evaluating Pillar 3)

Use these when checking standard alignment:

- **Standard A: Instructional Leadership for Continuous Improvement** — Focus on high-quality instruction, data-driven decisions, curriculum alignment, and continuous improvement processes
- **Standard B: Culture for Teaching and Learning** — Establishing a culture of high expectations, equity, and professional growth for all stakeholders
- **Standard C: Professional Learning and Growth** — Leading professional development, fostering collaborative learning communities, and supporting teacher growth
- **Standard D: Resource Management** — Strategic allocation of fiscal, human, and technological resources to support the school's mission

## Agent Teams Strategy

For comprehensive reviews, use agent teams to parallelize pillar evaluation:

### Team Setup

1. **Create the team**: `TeamCreate` with `team_name: "paper-review"`.
2. **Read the document first** to extract all paragraph text and identify section boundaries.
3. **Create 3 tasks** via `TaskCreate`:
   - **Content reviewer**: Evaluate Pillars 1-3 (Define & Contextualize, Research & Policy Review, Instructional Leadership Analysis). Check for synthesis vs. summary and TILS standard connections.
   - **Ethics reviewer**: Evaluate Pillars 4-5 (Risks & Ethical Considerations, Evidence-Based Recommendations). Check recommendations are actionable and research-grounded.
   - **Format reviewer**: Check APA formatting, typos, grammar, page length, and source count (minimum 5 scholarly sources).
4. **Spawn 3 teammates** via `Agent` tool with `team_name: "paper-review"`, names: `content-reviewer`, `ethics-reviewer`, `format-reviewer`. Use `subagent_type: "general-purpose"`. Launch all in a single message.
5. **Pass document content and paragraph indices** to each teammate so they can return precise locations.
6. Each teammate reviews its assigned areas and returns comments: `{para_index, search_text, comment_text, category}`.
7. **Coordinator merges all comments**, inserts into the document using the technique below, and saves.
8. **Shut down teammates** and `TeamDelete`.

## Technical Implementation

Use the same Word comment insertion technique as the `doc-review` skill:

```python
from docx import Document
from docx.oxml.ns import qn
from docx.oxml import OxmlElement
from lxml import etree
from docx.opc.part import Part
from docx.opc.packuri import PackURI

# Build comments XML container
comments_xml = '''<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<w:comments xmlns:wpc="http://schemas.microsoft.com/office/word/2010/wordprocessingCanvas"
            xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006"
            xmlns:o="urn:schemas-microsoft-com:office:office"
            xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships"
            xmlns:m="http://schemas.openxmlformats.org/officeDocument/2006/math"
            xmlns:v="urn:schemas-microsoft-com:vml"
            xmlns:wp="http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing"
            xmlns:w10="urn:schemas-microsoft-com:office:word"
            xmlns:w="http://schemas.openxmlformats.org/wordprocessingml/2006/main"
            xmlns:wne="http://schemas.microsoft.com/office/word/2006/wordml">
</w:comments>'''
comments_element = etree.fromstring(comments_xml.encode('utf-8'))

def add_comment_to_xml(comments_element, comment_id, author, date_str, text):
    comment = etree.SubElement(comments_element, qn('w:comment'))
    comment.set(qn('w:id'), str(comment_id))
    comment.set(qn('w:author'), author)
    comment.set(qn('w:date'), date_str)
    p = etree.SubElement(comment, qn('w:p'))
    r = etree.SubElement(p, qn('w:r'))
    t = etree.SubElement(r, qn('w:t'))
    t.text = text
    t.set(qn('xml:space'), 'preserve')

def add_comment_markers(paragraph, search_text, comment_id):
    found = False
    for run in paragraph.runs:
        if search_text.lower() in (run.text or "").lower():
            comment_start = OxmlElement('w:commentRangeStart')
            comment_start.set(qn('w:id'), str(comment_id))
            run.element.addprevious(comment_start)
            comment_end = OxmlElement('w:commentRangeEnd')
            comment_end.set(qn('w:id'), str(comment_id))
            run.element.addnext(comment_end)
            ref_run = OxmlElement('w:r')

From jasonbhorne/claude-code-skills, MIT.

More from jasonbhorne/claude-code-skills