Office PowerPoint MCP Server (office-powerpoint-mcp-server)

A production-ready Model Context Protocol (MCP) server for automated PowerPoint (.pptx) generation, OpenXML-enhanced financial styling, and presentation inspection. Built with Python 3.12+, FastMCP v4, python-pptx, lxml, Pillow, and Pydantic v2.

Inspired by and functionally referencing GongRzhe/Office-PowerPoint-MCP-Server, upgraded to modern 2026 standards with path sandboxing, consolidated tool surface, hybrid state persistence, and native OpenXML table formatting.

Key Architectural Highlights

Related MCP server: PPTX MCP Server

Installation

Install using uv:

# Clone or navigate to the project directory
cd office-powerpoint-mcp-server

# Create virtual environment and sync dependencies
uv sync

Running the MCP Server

The server provides a standard CLI entry point ppt-mcp-server:

1. Standard I/O (Default for MCP Clients like Claude Desktop, Cursor, Antigravity)

# Sandboxed to current working directory
uv run ppt-mcp-server

# Sandboxed to specific directories
uv run ppt-mcp-server stdio --allow-dir /Users/username/decks --allow-dir /Volumes/Presentations

2. HTTP Transport (Streamable HTTP / SSE)

uv run ppt-mcp-server http --host 127.0.0.1 --port 8000 --allow-dir /Users/username/decks

Client Configuration

Claude Desktop / Antigravity (claude_desktop_config.json / mcp-config.json)

{
  "mcpServers": {
    "powerpoint": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/path/to/office-powerpoint-mcp-server",
        "ppt-mcp-server",
        "stdio",
        "--allow-dir",
        "/path/to/allowed/decks"
      ]
    }
  }
}

Tool Reference (14 Consolidated Tools)

All tools expose MCP ToolAnnotations (read_only_hint, destructive_hint, idempotent_hint).

1. Lifecycle & Ingestion (tools/lifecycle.py)

• manage_presentation: Actions: create, open, save, close, set_properties.

  • Sets widescreen 16:9 (13.333" x 7.5") by default.

  • Supports hybrid in-memory IDs and direct disk paths with auto_save: true.

• inspect_presentation (Read-Only): Ingests an entire presentation in one call.

  • Returns slide counts, layout names, speaker notes, all text frames, and table rows as structured JSON or publication Markdown.

2. Themes & Institutional Styling (tools/design.py & templates.json)

• apply_theme: Applies institutional color schemes, slide background, and font hierarchies across slides or tables.

  • Supported Themes: institutional_navy, modern_blue, corporate_gray, elegant_green, warm_red.

• format_master_footer: Adds institutional divider line, confidentiality notice, current date, and dynamic page numbers (Slide X of Y) across slides.

3. Template-Driven Slide Generation (tools/slides.py)

• add_slide_from_template: Builds pre-calculated coordinate slides from templates.json:

  1. title_slide: Authoritative title slide with decorative accent bar, subtitle, author, and date.

  2. executive_summary: Dual strategic cards for observations and recommendations.

  3. two_column_comparison: Side-by-side comparison cards (Option A vs Option B / Current vs Target).

  4. kpi_metrics_grid: 4 high-impact metric cards with prominent numbers, labels, and YoY trends.

  5. table_with_commentary: Left-side financial data table with right-side commentary card.

  6. chart_with_takeaways: Left-side native chart with right-side analyst takeaways card.

• manage_slides: Slide operations: add, duplicate, delete, reorder, populate_placeholder.

• list_templates (Read-Only): Lists all available layout templates and color schemes.

4. Financial & Data Tables (tools/tables.py)

• add_data_table: Creates publication-ready financial tables:

  • Header row: Theme primary background (#1B365D) with bold white text.

  • Alternating rows: Slate gray zebra striping (#F1F5F9).

  • Subtle OpenXML horizontal borders (#CBD5E1) via lxml.

  • Left-aligned text in Column 0, right-aligned numbers in Columns $\ge 1$.

  • Optional bottom source / footnote line.

• update_table_cells: Fine-grained cell updates (text, colors, bold, alignment, font sizes).

5. Text & Shape Management (tools/content.py)

• manage_text: Add, update, or clear text boxes and multi-level bulleted lists (levels 0–4) with run-level overrides and auto-fit shrink-to-fit sizing.

• add_bullet_slide: Quick creation of institutional structured bullet slide with header and divider.

• manage_shapes: Create cards, callout containers, arrows, rectangles, and rounded containers.

6. Media & Charts (tools/media_charts.py)

• add_native_chart: Creates native editable PowerPoint charts (COLUMN_CLUSTERED, LINE, PIE, BAR_CLUSTERED) using CategoryChartData.

• add_image: Inserts images from a validated local file path or base64 string (e.g. from matplotlib), locking aspect ratio within a bounding box (left, top, max_width, max_height).

Institutional Navy Palette Specifications

Primary Accent / Header: '#1B365D' (Muted Institutional Navy)
Secondary Accent:        '#0F172A' (Slate Navy)
Zebra Row / Card Fill:   '#F1F5F9' (Slate Gray)
Border Lines:            '#CBD5E1' (Light Slate Border)
Body Text:               '#1A1A1A' (Dark Charcoal)
Muted Subtext:           '#64748B' (Muted Slate)
Background:              '#FFFFFF' (Pure White)
Fonts:
  Title: 'Helvetica'
  Body:  'Arial'

Quality & Automated Verification

Run tests and code formatting checks:

# Run full pytest test suite
uv run pytest -v

# Run Ruff linter and formatter checks
uv run ruff check .
uv run ruff format --check .

License

MIT License.