This PR introduces a comprehensive PowerPoint automation skill under the experimental collection, enabling AI-assisted creation, extraction, and validation of slide decks through a YAML-driven content pipeline. The implementation delivers a three-agent orchestration system, bidirectional PPTX-to-YAML conversion across 14 Python scripts, a PowerShell orchestrator, and dual-path validation combining PPTX property checks with Copilot SDK vision model analysis. Supporting changes include a new VS Code Playwright screenshot skill, collection manifest updates, Python virtual environment gitignore patterns, and a documentation file rename.

Description

PowerPoint Skill Architecture

The skill establishes a full-featured PPTX automation toolkit at .github/skills/experimental/powerpoint/, organized around a YAML-driven content model and bidirectional conversion pipeline.

• Added build_deck.py implementing two exclusive build modes: full rebuild from corporate templates (--template) and partial in-place regeneration of specific slides (--source + --slides). Uses an ELEMENT_BUILDERS registry dictionary for element dispatch supporting 11 element types (shape, textbox, rich_text, image, card, arrow_flow, numbered_step, table, chart, connector, group), z-order sorting, turbo mode for dense slides, and content-extra.py post-hooks for custom rendering.
• Added extract_content.py for reverse-engineering existing PPTX files into structured YAML with type-specific dispatch, global style detection via frequency analysis, light/dark slide classification, multi-theme generation, and @theme_name preservation.
• Added export_slides.py and render_pdf_images.py for PPTX-to-PDF conversion via LibreOffice with platform-specific executable detection, optional slide filtering, and PDF-to-JPG rasterization with original slide number mapping.
• Added validate_deck.py for offline PPTX property validation (speaker notes, slide count) with per-slide JSON output, and validate_slides.py for vision model-based quality analysis via Copilot SDK with exponential backoff retry.

Utility Module Library

Shared patterns were consolidated into modular utilities following a bidirectional apply/extract design for PPTX-to-YAML round-trip fidelity.

• Added pptx_text.py with three populate modes (flat, per-paragraph, rich-text), key-mapping abstraction for element-type-specific YAML conventions, and comprehensive paragraph/run property support including bullets, hyperlinks, and character spacing.
• Added pptx_fills.py for solid, gradient, and pattern fills with alpha transparency via XML alphaModFix manipulation, line properties, and outer shadow effects via effectLst.
• Added pptx_charts.py supporting 12 chart types, pptx_tables.py with cell merging and banding, pptx_colors.py for hex/theme/brightness color resolution, pptx_fonts.py for font normalization, pptx_shapes.py with 30+ shape type mappings, and pptx_utils.py centralizing exit codes, logging, EMU conversion, and slide filter parsing.

Agent and Prompt System

• Added pptx.agent.md as the PowerPoint Builder orchestrator coordinating five phases: research, build-content, build-deck, validate, and export. Delegates all execution to subagents while maintaining coordination artifacts.
• Added pptx-subagent.agent.md executing delegated tasks (extract, build-content, build-deck, validate, export) with a blocking failure protocol that prevents silent degradation on unexpected results.
• Added pptx.instructions.md defining shared conventions for working directory structure, content schemas, validation criteria, and build modes.
• Added pptx.prompt.md as the user-facing entry point for initiating PowerPoint workflows.

Skill Documentation and Configuration

• Added comprehensive SKILL.md covering prerequisites, content directory structure, YAML schemas, script reference, architecture, constraints, validation rules, and troubleshooting.
• Added content-yaml-template.md (complete YAML schema for slide content), content-extra-py-template.md (custom Python rendering), and style-yaml-template.md (dimensions, layout mapping, metadata, theme defaults).
• Added pyproject.toml declaring dependencies (python-pptx, pyyaml, cairosvg, Pillow, pymupdf, github-copilot-sdk) with dev dependencies for pytest and ruff.
• Added Invoke-PptxPipeline.ps1 as the PowerShell orchestrator managing virtual environment lifecycle via uv sync, platform-specific Python path resolution, and delegation to Python scripts.

Test Suite

• Added 14 Python test modules and 1 PowerShell Pester test suite with shared conftest.py fixtures. Coverage spans all element types, round-trip fidelity, platform-specific behavior, and mocked external dependencies (LibreOffice, PyMuPDF, Copilot SDK).
  • test_build_deck.py (1,270 lines) covers all 11 element types, z-order, turbo mode, cards, arrow flows, numbered steps, and groups
  • test_extract_content.py (1,169 lines) covers extraction, theme resolution, and global style detection
  • test_pptx_text.py (~800 lines) covers text frame operations, bullets, and rich-text formatting
  • Remaining modules verify charts, colors, fills, fonts, shapes, tables, utils, PDF rendering, and validation
  • Invoke-PptxPipeline.Tests.ps1 (413 lines) validates pipeline orchestration with platform-specific handling
• Added uv.lock pinning 30+ dependency versions.

VS Code Playwright Screenshot Skill

• Added .github/skills/experimental/vscode-playwright/SKILL.md providing a workflow for capturing VS Code screenshots via Playwright MCP tools with serve-web, covering server lifecycle, viewport configuration, UI cleanup, theme seeding, and screenshot validation.

Infrastructure and Documentation

• Updated .gitignore with venv/ and .venv/ patterns for Python virtual environments.
• Updated collections/experimental.collection.yml with the powerpoint skill entry.
• Regenerated plugins/experimental/README.md and plugins/hve-core-all/README.md with the powerpoint skill row.
• Renamed docs/customization/agents.md to docs/customization/custom-agents.md and updated the corresponding link in docs/README.md.

Related Issue(s)

None

Type of Change

Select all that apply:

Code & Documentation:

• Bug fix (non-breaking change fixing an issue)
• New feature (non-breaking change adding functionality)
• Breaking change (fix or feature causing existing functionality to change)
• Documentation update

Infrastructure & Configuration:

• GitHub Actions workflow
• Linting configuration (markdown, PowerShell, etc.)
• Security configuration
• DevContainer configuration
• Dependency update

AI Artifacts:

• Reviewed contribution with prompt-builder agent and addressed all feedback
• Copilot instructions (.github/instructions/*.instructions.md)
• Copilot prompt (.github/prompts/*.prompt.md)
• Copilot agent (.github/agents/*.agent.md)
• Copilot skill (.github/skills/*/SKILL.md)

Note for AI Artifact Contributors:

• Agents: Research, indexing/referencing other project (using standard VS Code GitHub Copilot/MCP tools), planning, and general implementation agents likely already exist. Review .github/agents/ before creating new ones.
• Skills: Must include both bash and PowerShell scripts. See Skills.
• Model Versions: Only contributions targeting the latest Anthropic and OpenAI models will be accepted. Older model versions (e.g., GPT-3.5, Claude 3) will be rejected.
• See Agents Not Accepted and Model Version Requirements.

Other:

• Script/automation (.ps1, .sh, .py)
• Other (please describe):

Sample Prompts (for AI Artifact Contributions)

User Request:

Invoke the PowerPoint Builder workflow using @pptx build me a slide deck about cloud-native architecture for a customer presentation. The prompt supports topic-based generation, extraction from existing PPTX files, and iterative validation with vision model feedback.

Execution Flow:

1. The orchestrator agent (pptx.agent.md) receives the request and delegates research to the Researcher Subagent for topic investigation and content gathering.
2. The PowerPoint Subagent executes build-content to generate per-slide YAML content files from research findings.
3. The subagent runs build-deck via Invoke-PptxPipeline.ps1, which manages the Python virtual environment and calls build_deck.py to produce the PPTX file.
4. The subagent runs validate to check PPTX properties (speaker notes, slide count) and optionally sends rendered images to the Copilot SDK vision model for quality analysis.
5. Validation findings drive iteration back to content fixes or design adjustments (up to 5 iterations).
6. The subagent runs export to render final slide images for visual review.

Output Artifacts:

• .copilot-tracking/ppt/{session}/content/slide-NNN/content.yaml — Per-slide YAML definitions
• .copilot-tracking/ppt/{session}/content/style.yaml — Global style configuration
• .copilot-tracking/ppt/{session}/output/*.pptx — Generated PowerPoint deck
• .copilot-tracking/ppt/{session}/output/slide-NNN.jpg — Rendered slide images
• .copilot-tracking/ppt/{session}/validation/report.md — Validation findings

Success Indicators:

• The generated PPTX opens in PowerPoint or LibreOffice and renders all slides with correct layout, styling, and content. Validation reports show no errors. Speaker notes are present on all slides. Slide count matches the content directory structure.

Testing

• Automated validation commands run during PR generation (see Checklist below).
• Security analysis found no sensitive data exposure, dependency vulnerabilities, or privilege escalation concerns.
• Diff-based assessment confirmed all 46 changed files follow repository naming conventions.
• Manual testing was not performed during PR generation.

Checklist

Required Checks

• Documentation is updated (if applicable)
• Files follow existing naming conventions
• Changes are backwards compatible (if applicable) (N/A — all files are new additions)
• Tests added for new functionality (if applicable)

AI Artifact Contributions

• Used /prompt-analyze to review contribution
• Addressed all feedback from prompt-builder review
• Verified contribution follows common standards and type-specific requirements

Required Automated Checks

The following validation commands must pass before merging:

• Markdown linting: npm run lint:md
• Spell checking: npm run spell-check
• Frontmatter validation: npm run lint:frontmatter
• Skill structure validation: npm run validate:skills
• Link validation: npm run lint:md-links
• PowerShell analysis: npm run lint:ps
• Plugin freshness: npm run plugin:generate

GHCP Artifact Maturity

GHCP Maturity Acknowledgment

• I acknowledge this PR includes non-stable GHCP artifacts
• Non-stable artifacts are intentional for this change

Security Considerations

• This PR does not contain any sensitive or NDA information
• Any new dependencies have been reviewed for security issues
• Security-related scripts follow the principle of least privilege

Additional Notes

• The vscode-playwright/SKILL.md was added to the repository but is not yet listed in the experimental.collection.yml manifest. Plugin generation auto-added it to hve-core-all.collection.yml; consider also adding it to the experimental collection for consistency.
• Added invoke-pptx-pipeline.sh bash script to satisfy the skill validation requirement for a .sh file in scripts/.
• Added BOM encoding to Invoke-PptxPipeline.ps1 to resolve PSScriptAnalyzer PSUseBOMForUnicodeEncodedFile warning.
• The plugins/experimental/README.md row for powerpoint shows powerpoint | powerpoint with duplicate description text — this is generated output and may warrant updating the collection metadata to include a proper description.
• All 31 commits follow conventional commits format.