Agent Skills: PowerPoint Generation (pptx)

PowerPoint Generation (pptx)

Overview

Generate .pptx files programmatically using python-pptx. Convert Claude output, data analysis results, research reports, and structured data into polished PowerPoint presentations.

When to Invoke

Skill({ skill: 'pptx' }) when:

• User asks to "create a presentation", "generate slides", or "make a PowerPoint"
• Converting a report, analysis, or structured document into slide format
• Generating recurring report slides from data
• Creating template-based presentations programmatically

Installation

pip install python-pptx
# or
uv add python-pptx

Quick Start

from pptx import Presentation
from pptx.util import Inches, Pt, Emu
from pptx.enum.text import PP_ALIGN
from pptx.dml.color import RGBColor

# Create presentation
prs = Presentation()

# Set slide dimensions (widescreen 16:9)
prs.slide_width = Inches(13.33)
prs.slide_height = Inches(7.5)

# Add title slide using layout 0
title_layout = prs.slide_layouts[0]
slide = prs.slides.add_slide(title_layout)
title = slide.shapes.title
subtitle = slide.placeholders[1]

title.text = "Q1 2026 Business Review"
subtitle.text = "Prepared by Claude Agent"

# Add content slide using layout 1 (title + content)
content_layout = prs.slide_layouts[1]
slide = prs.slides.add_slide(content_layout)
slide.shapes.title.text = "Key Findings"

tf = slide.placeholders[1].text_frame
tf.text = "Revenue grew 23% YoY"
p = tf.add_paragraph()
p.text = "Customer retention at 94%"
p.level = 1  # Indent level

prs.save("presentation.pptx")

Slide Layouts Reference

Standard layouts available in default template:

| Index | Name | Use For | | ----- | --------------------- | --------------------------- | | 0 | Title Slide | Opening slide | | 1 | Title and Content | Main content slides | | 2 | Title and Two Content | Side-by-side comparison | | 5 | Title Only | Charts or images with title | | 6 | Blank | Custom layouts |

# List all layouts in a template
for i, layout in enumerate(prs.slide_layouts):
    print(f"{i}: {layout.name}")

Text Formatting

from pptx.util import Pt
from pptx.dml.color import RGBColor
from pptx.enum.text import PP_ALIGN

def format_paragraph(paragraph, text, size=18, bold=False, color=None, align=PP_ALIGN.LEFT):
    run = paragraph.add_run()
    run.text = text
    font = run.font
    font.size = Pt(size)
    font.bold = bold
    if color:
        font.color.rgb = RGBColor(*color)  # e.g., (0x1F, 0x49, 0x7D) for dark blue
    paragraph.alignment = align
    return run

# Usage
tf = slide.placeholders[1].text_frame
tf.clear()  # Clear existing content
p = tf.paragraphs[0]
format_paragraph(p, "Important Metric: 94%", size=24, bold=True, color=(0x1F, 0x49, 0x7D))

Tables

from pptx.util import Inches

def add_table_slide(prs, title, headers, rows):
    """Add a slide with a formatted table."""
    slide = prs.slides.add_slide(prs.slide_layouts[5])  # Title Only
    slide.shapes.title.text = title

    # Position: left, top, width, height
    left = Inches(0.5)
    top = Inches(1.5)
    width = Inches(12.3)
    height = Inches(0.8 * (len(rows) + 1))

    table = slide.shapes.add_table(
        len(rows) + 1, len(headers), left, top, width, height
    ).table

    # Header row
    for col_idx, header in enumerate(headers):
        cell = table.cell(0, col_idx)
        cell.text = header
        cell.text_frame.paragraphs[0].runs[0].font.bold = True

    # Data rows
    for row_idx, row in enumerate(rows):
        for col_idx, value in enumerate(row):
            table.cell(row_idx + 1, col_idx).text = str(value)

    return slide

# Usage
headers = ["Product", "Revenue", "Growth"]
rows = [
    ("Widget A", "$1.2M", "+23%"),
    ("Widget B", "$890K", "+15%"),
    ("Widget C", "$450K", "-5%"),
]
add_table_slide(prs, "Revenue by Product", headers, rows)

Charts

from pptx.chart.data import ChartData
from pptx.enum.chart import XL_CHART_TYPE
from pptx.util import Inches

def add_bar_chart(slide, title, categories, series_name, values):
    """Add a bar chart to a slide."""
    chart_data = ChartData()
    chart_data.categories = categories
    chart_data.add_series(series_name, values)

    chart = slide.shapes.add_chart(
        XL_CHART_TYPE.COLUMN_CLUSTERED,
        Inches(1), Inches(1.5),   # left, top
        Inches(11), Inches(5.5),  # width, height
        chart_data
    ).chart

    chart.has_title = True
    chart.chart_title.text_frame.text = title
    chart.has_legend = False

    return chart

# Usage
slide = prs.slides.add_slide(prs.slide_layouts[5])
slide.shapes.title.text = "Monthly Revenue"
add_bar_chart(
    slide,
    "Revenue by Month (2026)",
    ["Jan", "Feb", "Mar"],
    "Revenue ($K)",
    (450, 520, 610)
)

Images

from pptx.util import Inches

def add_image_slide(prs, title, image_path, caption=None):
    slide = prs.slides.add_slide(prs.slide_layouts[5])
    slide.shapes.title.text = title

    # Center image on slide
    pic = slide.shapes.add_picture(
        image_path,
        left=Inches(1.5),
        top=Inches(1.5),
        width=Inches(10),
    )

    if caption:
        txBox = slide.shapes.add_textbox(
            Inches(1.5), Inches(6.8), Inches(10), Inches(0.5)
        )
        tf = txBox.text_frame
        tf.text = caption
        tf.paragraphs[0].alignment = PP_ALIGN.CENTER

    return slide

Agent Workflow: Claude Output → Slides

import json
from pptx import Presentation
from pptx.util import Inches, Pt

def claude_output_to_slides(slide_data: list[dict], output_path: str):
    """
    Convert Claude structured output to a PowerPoint file.

    Expected slide_data format:
    [
      {"type": "title", "title": "...", "subtitle": "..."},
      {"type": "content", "title": "...", "bullets": ["...", "..."]},
      {"type": "table", "title": "...", "headers": [...], "rows": [[...], ...]},
    ]
    """
    prs = Presentation()
    prs.slide_width = Inches(13.33)
    prs.slide_height = Inches(7.5)

    for slide_spec in slide_data:
        if slide_spec["type"] == "title":
            slide = prs.slides.add_slide(prs.slide_layouts[0])
            slide.shapes.title.text = slide_spec["title"]
            if "subtitle" in slide_spec:
                slide.placeholders[1].text = slide_spec["subtitle"]

        elif slide_spec["type"] == "content":
            slide = prs.slides.add_slide(prs.slide_layouts[1])
            slide.shapes.title.text = slide_spec["title"]
            tf = slide.placeholders[1].text_frame
            tf.text = slide_spec["bullets"][0]
            for bullet in slide_spec["bullets"][1:]:
                p = tf.add_paragraph()
                p.text = bullet
                p.level = 1

        elif slide_spec["type"] == "table":
            add_table_slide(
                prs,
                slide_spec["title"],
                slide_spec["headers"],
                slide_spec["rows"]
            )

    prs.save(output_path)
    return output_path

# Example: Generate from Claude JSON output
slide_data = [
    {"type": "title", "title": "Q1 2026 Report", "subtitle": "Generated by Claude"},
    {"type": "content", "title": "Key Highlights", "bullets": [
        "Revenue up 23% YoY",
        "New customer acquisition: +450",
        "NPS score improved to 72",
    ]},
    {"type": "table", "title": "Regional Performance",
     "headers": ["Region", "Revenue", "Target", "Variance"],
     "rows": [["North", "$2.1M", "$2.0M", "+5%"], ["South", "$1.8M", "$2.0M", "-10%"]]},
]
claude_output_to_slides(slide_data, "q1-report.pptx")

Using Existing Templates

# Load existing branded template
prs = Presentation("company-template.pptx")

# Add slide using template's layouts
# Template layouts preserve branding (fonts, colors, logos)
slide = prs.slides.add_slide(prs.slide_layouts[1])
slide.shapes.title.text = "New Content Slide"

Output Location

Save generated .pptx files to:

• User-specified path, or
• .claude/context/artifacts/ for agent-generated presentations

Anti-Patterns

• Never use raw integer values for sizes — always Pt(), Inches(), or Emu()
• Never add text directly to slide (use placeholder shapes for layout consistency)
• Never hardcode colors without using RGBColor — inconsistent on dark/light themes
• Never forget tf.clear() before setting text on a reused placeholder
• Never use slide_layouts[6] (blank) for content — placeholder positions will be missing

Related Skills

• xlsx — Excel spreadsheet generation (tabular data export)
• diagram-generator — Mermaid diagrams (for embedding architecture diagrams)
• markitdown-converter — Convert existing files to Markdown for Claude processing

Memory Protocol (MANDATORY)

Before starting: Read .claude/context/memory/learnings.md

After completing: Record any python-pptx version compatibility issues or slide layout gotchas.

ASSUME INTERRUPTION: Your context may reset. If it's not in memory, it didn't happen.