Story-to-Slides Skill

Purpose

Read any narrative document with an existing story arc and produce an optimized presentation brief that the PPTX skill can render into slides. You are a presentation strategist: analyze the narrative's argument structure, distill it into slide-level messages using pyramid communication, apply copywriting techniques, and select the right visual layout for each message.

A great presentation brief is not a transcript of the narrative. It is a re-architecture of the narrative's argument into a visual medium where every slide has ONE clear message, supported by evidence the audience can absorb in 3 seconds. Slides that try to convey multiple messages become walls of text that audiences tune out — the presenter loses control of the room.

Architecture

Two-layer intelligence:

1. Story Arc Analysis — read narrative, identify argument structure, extract governing thought, map section roles
2. Message Architecture + Slide Specification — pyramid communication, one message per slide, copywriting, layout selection, speaker notes

The brief describes WHAT each slide says and which layout to use. The PPTX renderer owns all visual decisions (colors, fonts, spacing) by reading the theme directly — briefs contain no color fields.

Density principle: Slides carry the anchor; speaker notes carry the detail. A McKinsey partner's slide has one assertion headline and 3-5 scannable phrases — the partner delivers the depth from memory. Same principle here: when content exceeds a layout's physical capacity, the excess moves to speaker notes. Never force-fit paragraphs on-slide.

Parameters

Parameter	Default	Description
source_path	auto-discovered	Narrative file or directory. When omitted with interactive=true, Step 0 searches nearby.
theme	interactive	Absolute path to theme.md, or omit to trigger cogni-workspace:pick-theme interactive selection.
language	en	Language code (en/de)
title / subtitle	auto-detected	Extracted from narrative if not provided
customer_name / provider_name	from metadata	Organization names
output_path	{source_dir}/cogni-visual/presentation-brief.md	Brief output location
max_slides	15	Maximum slide count (forces consolidation if narrative is long)
arc_type	auto	Story arc hint: why-change, problem-solution, journey, argument, report
arc_id	from frontmatter	Narrative arc ID from cogni-narrative (e.g., industry-transformation). Mapped to visual arc_type in Step 1.
arc_definition_path	none	Path to arc definition file — element names become methodology slide phase labels.
interactive	true	When true, present choices via AskUserQuestion. When false, auto-select.
stakeholder_review	interactive	When true, run brief-review-assessor after validation. Defaults to value of interactive.
audience_context	none	Structured audience/buyer data for targeted evidence selection and Q&A prep (Rich mode).

Caller-supplied overrides

These are typically set by an upstream agent (e.g., why-change-work), not by a human user:

Parameter	Default	Description
confidence_threshold	0.8	Minimum confidence for automatic layout mapping
governing_thought	auto-extracted	Pre-computed governing thought — Step 4 validates rather than re-derives.
section_roles	auto-detected	Pre-mapped section roles — Step 4 validates rather than re-derives.
buyer_appendix_path	none	Path to buyer-appendix.md for enriched Q&A prep (Step 8.2 only).

---

Conventions

Theme-driven visuals

The PPTX renderer owns all visual decisions — colors, fonts, spacing — by reading the theme directly. Briefs specify content and layout only. Omit visual fields (Background:, Text-Color:, Icon-Color:, Role:, Intensity:, Mood:) because the renderer ignores them and their presence creates ambiguity about who controls styling.

Interactive checkpoints

Interactive prompts let the user steer creative decisions at two points: narrative selection (Step 0) and theme selection (Step 1). Use the structured format below — unstructured prose renders as empty prompts:

questions: [{
  question: "Your question here?",
  header: "Short Label",
  options: [
    { label: "Option Name", description: "What this means" },
    { label: "Another Option", description: "What this means" }
  ],
  multiSelect: false
}]

On empty or blank responses, auto-select the best option and move on. When interactive is false, skip all AskUserQuestion calls.

German fidelity

German presentations go to executives. ASCII-ified umlauts (ae/oe/ue) immediately signal "machine-generated" and undermine credibility. Use real Unicode throughout: ae→ä oe→ö ue→ü ss→ß. German number formatting: 2.661 (dot as thousands separator).

Client-facing copy hygiene

Presentation briefs are client-facing deliverables. Strip all internal sales methodology vocabulary from slide titles, headlines, and body copy:

• Never expose: "Power Position", "Why Change", "Why Now", "Why You", "Why Pay", "Unconsidered Need", "Buying Center" in any client-visible field (Slide-Title, headlines, bullets, Bottom-Banner)
• Transform to: customer-benefit language that asserts a specific capability the customer cares about.
  • Bad: "Why You — Power Position 2 & 3: Compliance & Sovereign Cloud"
  • Good: "Warum T-Systems — Compliance-native Architektur eliminiert 90% der Audit-Findings"
• One Power Position = one slide. Never combine multiple Power Positions into a single slide title or body. Each gets its own slide with its own assertion headline.
• Internal-only slides (Methodology, Buying Center) are exempt — they carry the INTERNAL warning banner.

IS/DOES/MEANS label localization

The is-does-means layout uses layer labels rendered as badges on each box. These labels must match the presentation language:

Language	IS label	DOES label	MEANS label
en	IS	DOES	MEANS
de	IST	MACHT	BEDEUTET

Apply the localized labels in every IS-Box.Label, DOES-Box.Label, MEANS-Box.Label field. English labels in a German presentation signal "template not adapted" and undermine credibility.

---

Quick Reference: Good vs Bad Output

Headlines — assert, don't label:

• Bad: "Market Overview" (topic label — audience must read the body to get the point)
• Good: "European Rail Market Contracts 12% as Regulation Tightens" (assertion — message lands in 3 seconds)

Number plays — reframe for impact:

• Bad: "There were 688 rail suicides in 2023" (buried in prose)
• Good: Hero number 688 isolated in stat-card, sublabel + 2,661 station attacks, context box explains why manual monitoring fails

Bullets — scan-optimized, not sentences:

• Bad: "The security staff are unable to adequately cover all areas of the rail network on a 24/7 basis" (19 words)
• Good: "Staff cannot cover all areas 24/7" (6 words — scannable in ~3 seconds at presentation distance, ~8-10 words max)

---

Slide Structure

Position	Layout	Role
First	title-slide	Opening with title, subtitle, metadata
After title	1-2 internal prep slides	Methodology (always) + Buying Center (Rich mode only)
Body	Content slides	One message per slide
Second-to-last	closing-slide	CTA headline, key takeaway
Last	References slide	Consolidated citations

Internal prep slides carry Bottom-Banner with "INTERNAL — REMOVE FROM CLIENT PRESENTATION" and are not counted against max_slides.

---

Workflow

When invoked without explicit parameters, search the filesystem first (Step 0) rather than prompting for paths. Users invoke this skill from project directories that already contain their narrative — asking for a path they're sitting next to creates unnecessary friction. The two interactive checkpoints are: (1) narrative selection in Step 0, and (2) theme selection in Step 1 via cogni-workspace:pick-theme.

Execution protocol

Each step: verify the previous step's output is available (entry gate), read the reference file for that step, execute, then state your output summary before moving on. Reference files contain step-specific rules that prevent downstream rework — read them at the start of each step.

---

Step 0: Narrative Auto-Discovery

Users invoke from project directories containing their narrative. Searching first eliminates path-fumbling.

If source_path was explicitly provided: set source_dir to its parent directory and proceed to Step 1.

Otherwise, search without asking:

1. Primary: Glob **/insight-summary.md from CWD (max 3 levels)
2. For each candidate: read first 30 lines, extract title, arc_id, estimate word count
3. Secondary (if 0 primary results): Glob **/*.md, filter for arc_id: in first 30 lines. Exclude SKILL.md, README.md, CLAUDE.md.
4. Sort: insight-summary.md files first, then by path depth (shallow first)

If candidates found: Present via AskUserQuestion (max 4 options with filename, title, arc_id, word count). On empty response, auto-select top candidate.

If no candidates: Ask for path or cancel. On empty response, stop with: "No narrative path provided. Stopping."

Set source_dir = parent directory of selected source_path.

---

Step 1: Parse Parameters & Resolve Context

Arc resolution and theme loading happen before reading the narrative because they shape how you interpret the story — a pre-resolved arc_type tells you what argument pattern to look for.

Determine input type (directory with metadata vs single file) and load metadata.

Arc resolution (priority order):

1. arc_id parameter → use directly
2. Source narrative frontmatter arc_id → extract
3. Neither → Step 4 auto-detects

If arc_id set: read $CLAUDE_PLUGIN_ROOT/libraries/arc-taxonomy.md, map to arc_type. If arc_definition_path provided: extract element names for methodology slide.

Theme resolution: Delegate to cogni-workspace:pick-theme — the ecosystem-standard theme picker.

1. If theme parameter was explicitly provided with an absolute path: use it directly, skip the picker.
2. Otherwise: invoke the cogni-workspace:pick-theme skill via the Skill tool. The picker scans standard and workspace theme directories, presents an interactive AskUserQuestion, and returns the absolute theme_path.
3. Store the returned theme_path (absolute path to theme.md), theme_name, and theme_slug for downstream use.
4. Read the selected theme.md to confirm it loads correctly.

If pick-theme is unavailable (e.g., cogni-workspace not installed), fall back to Glob scanning $COGNI_WORKSPACE_ROOT/themes/*/theme.md and present via AskUserQuestion manually.

Load libraries: cta-taxonomy.md, arc-taxonomy.md (if arc_id set). The heavier libraries (pptx-layouts.md and EXAMPLE_BRIEF.md) are deferred to the steps that consume them (Steps 7 and 8) to keep context lean during the creative intelligence steps.

---

Step 2: Read Narrative Content

Citation renumbering feeds 6 downstream steps — errors here propagate everywhere.

Read all source files, preserve section boundaries (H1/H2/H3), extract citation URL map, assign sequential citation numbers. Store renumber map for Steps 6, 8.1, 10.

Citation rules: See references/07-output-template.md (Citation Handling Rules section) for preservation rules, exclusion zones, and Source field generation priority.

Step 2.1: Diagram Detection

Read references/2g-diagram-simplification.md. Scan for Mermaid fenced blocks, classify type (gantt/layered-architecture/process-flow), check complexity against layout constraints, simplify if needed. IS-DOES-MEANS slides are not affected by diagram detection.

---

Step 3: Build Audience Model

The audience model shapes how downstream steps select evidence, frame the governing thought, and prepare Q&A — Rich mode enables targeted speaker notes, Lean mode infers from narrative vocabulary.

Read reference: references/02-audience-model.md

Build Audience Model: Rich mode (from audience_context or pitch-log.json) or Lean mode (inferred from narrative). Identify primary decision-maker, priorities, objections.

Content checkpoint: State mode, confidence, decision-maker, top priority, top objection.

---

Step 4: Analyze Story Arc

The governing thought and arc type cascade through everything downstream — message architecture, consolidation, layout selection, and speaker notes coaching. Getting them right here prevents rework in later steps.

Read reference: references/03-story-arc-analysis.md

• 4a. Detect arc type — why-change, problem-solution, journey, argument, or report
• 4b. Extract governing thought — single sentence, audience-weighted
• 4c. Map section roles — hook, problem, urgency, evidence, solution, proof, options, roadmap, investment, call-to-action

When caller provides governing_thought/section_roles: validate against narrative rather than re-deriving. When arc_context populated from Step 1: use resolved arc_type directly.

Content checkpoint: State arc type, governing thought, sections mapped count.

---

Step 5: Architect Slide Messages

Pyramid communication separates a professional deck from a content dump. Without one-message-per-slide discipline, the natural tendency is to pack 3-4 points per slide — and the audience remembers none of them.

Read reference: references/04-message-architecture.md

• 5a. Pyramid Principle — governing thought → arguments → evidence
• 5b. One message per slide — each slide gets exactly ONE message sentence
• 5c. Consolidation — merge/cut when narrative exceeds max_slides. Rich mode protects slides aligned with decision-maker priorities.
• 5d. MECE check — mutually exclusive, collectively exhaustive, logically ordered

Content checkpoint: State slide count, argument count, consolidation status.

---

Step 6: Apply Copywriting Techniques

Headlines are the first thing an audience reads. A topic label ("Market Overview") tells them nothing — they have to read the body to get the point, and you've lost their attention. An assertion headline ("European market contracts 12% as regulation tightens") delivers the message in 3 seconds.

Read reference: references/05a-slide-copywriting.md

• 6a. Headline optimization — every title is a complete action assertion (up to ~100 chars, contains verb + quantified consequence)
• 6b. Number plays — ratio framing, hero number isolation, before/after contrast
• 6c. Bullet consolidation — 8-12 points → 3-5 scannable bullets (max 10 words each)
• 6d. Evidence selection — top 3-5 per slide, audience-weighted, inline citations preserved

Speaker-Notes are generated later in Step 8.2 (after layouts finalize). This step focuses on slide copy only.

---

Step 6.1: Propose CTAs

Without explicit CTAs, the audience leaves impressed but without a next step. CTAs convert attention into action.

Read reference: $CLAUDE_PLUGIN_ROOT/libraries/cta-taxonomy.md

1. Extract implicit CTAs from narrative + generate from governing thought, arc type, hero numbers
2. Per content slide: assign cta.text (max 50 chars, imperative verb), cta.type (explore/evaluate/commit/share), cta.urgency
3. Build cta_summary: 3-5 proposals, primary_cta = highest-urgency commit CTA

If interactive: present CTA plan via AskUserQuestion (Approve/Adjust). On empty response, treat as approval.

---

Step 7: Select Layouts

Layout selection translates message type into visual structure. A hero number on a two-columns-equal wastes its impact — stat-card-with-context isolates the number and makes it the focal point.

Read library: $CLAUDE_PLUGIN_ROOT/libraries/pptx-layouts.md — slide layout schemas needed for layout selection (this step) and YAML field names (Step 8). Deferred from Step 1 to keep context lean during creative Steps 2-6.

Read reference: references/06-slide-mapping-rules.md

Map each slide to best layout from pptx-layouts.md. Mandatory rules apply first (title-slide, closing-slide), then diagram rules (Mermaid blocks), then content pattern matching. Ensure layout variety and confidence >= threshold.

---

Step 7.5: Density Pass — Compress to Layout Limits

Steps 5-6 optimized for message clarity. Now that layouts are assigned, compress each slide's copy to fit its layout's physical box dimensions. Speaker notes absorb the overflow. Think of this step as the difference between a draft memo and a billboard campaign — same message, radically different word budget.

Apply the layout density budget (HARD LIMITS):

Layout	Field	Max Words	Think of it as...
is-does-means	IS-Box	15	Conference badge tagline
is-does-means	DOES-Box	20	McKinsey "so what" bullet
is-does-means	MEANS-Box	15	Résumé skills line
stat-card	Context-Box Bullets (each)	10	Dashboard KPI label
four-quadrants	Bullets (each)	10	McKinsey slide bullet
two-columns-equal	Bullets (each)	10	McKinsey slide bullet
ALL	Bottom-Banner	12	Billboard tagline

Content exceeding the budget moves to speaker notes (Step 8.2 incorporates it) — it is preserved, not deleted. The slide carries the anchor; the presenter delivers the detail.

Why this works: The audience scans each slide in ~3 seconds before the presenter speaks. A 40-word IS-box becomes a reading competition — they read instead of listen, and the presenter loses the room. A 15-word phrase lets the audience absorb the anchor and look up, ready for the presenter's elaboration. The detail lives in speaker notes, not lost.

Content checkpoint: State slides compressed count, total words moved to notes.

---

Step 8: Generate YAML Slide Specifications

The YAML specification is the contract between this skill and the PPTX renderer. Every field must contain final, copy-paste-ready text because the renderer reproduces it exactly — no interpretation, no cleanup.

Read library: $CLAUDE_PLUGIN_ROOT/libraries/EXAMPLE_BRIEF.md — output format reference needed for YAML spec generation and final brief output (Step 10). Deferred from Step 1 to keep context lean during creative Steps 2-6.

Read reference: references/07-output-template.md (Slide YAML Example section)

For each slide, generate content-only YAML following pptx-layouts.md field names. Omit all visual fields — the renderer reads the theme. Every slide heading is an assertion headline.

Density enforcement (Step 7.5): IS/DOES/MEANS boxes: billboard-line brevity (15/20/15 words max, phrase notation only). All bullets: McKinsey slide bullet density (max 10 words, phrase not sentence). No full sentences in any box or bullet field.

---

Step 8.1: Generate References Slide

A consolidated references slide with working links gives the presentation credibility and lets readers verify claims independently.

Read reference: references/08b-references-slide.md

Generate from citation renumber map (Step 2). Position after closing-slide as last slide in deck.

---

Step 8.2: Enrich and Write Complete Brief (Delegated)

Speaker notes transform a deck from a document into a performance tool. Prep slides give the presenter strategic context before entering the room. The slides-enrichment-artist agent loads its own heavy references (1,647 lines) in a separate context, generates prep slides and speaker notes, and writes the complete presentation-brief.md directly — eliminating the token-heavy JSON round-trip and the integration step that can stall the orchestrator.

Prepare the enrichment prompt with these fields from previous steps:

Field	Source
OUTPUT_PATH	Resolved output path (from parameters or default {source_dir}/cogni-visual/presentation-brief.md)
OUTPUT_TEMPLATE_PATH	$CLAUDE_PLUGIN_ROOT/skills/story-to-slides/references/07-output-template.md
FRONTMATTER	All YAML frontmatter fields: type, version (4.0), theme, theme_path, customer, provider, language, generated, arc_type, arc_id, governing_thought, confidence_score, transformation_notes
TITLE / SUBTITLE	From Step 2 or parameters
SLIDE_SPECS	All slide YAML specs from Steps 8 + 8.1 (the complete deck so far)
AUDIENCE_MODEL	From Step 3 — mode (Rich/Lean), decision-maker, priorities, objections, champion, blockers
ARC_ANALYSIS	From Step 4 — arc_type, governing_thought, section_roles, arc phases
LANGUAGE	en or de
ARC_ID	If set
ARC_DEFINITION_PATH	If set (element names for methodology slide phase labels)
BUYER_APPENDIX_PATH	If set (enriched Q&A prep)
CTA_SUMMARY	From Step 6.1 (or "none")
GENERATION_METADATA_STATS	Raw stats: number_plays, headlines_optimized, bullets_consolidated, source_links, layout_distribution, avg_confidence, manual_review

Resolve output path before launching (run via Bash):

• If output_path explicit: mkdir -p "$(dirname "${output_path}")"
• Otherwise: set output_path = {source_dir}/cogni-visual/presentation-brief.md and mkdir -p "{source_dir}/cogni-visual"

Launch the slides-enrichment-artist agent:

Agent tool:
  subagent_type: "cogni-visual:slides-enrichment-artist"
  prompt: |
    OUTPUT_PATH: {resolved_output_path}
    OUTPUT_TEMPLATE_PATH: $CLAUDE_PLUGIN_ROOT/skills/story-to-slides/references/07-output-template.md

    FRONTMATTER:
      type: presentation-brief
      version: "4.0"
      theme: {theme_id}
      theme_path: "{theme_path}"
      customer: "{customer_name}"
      provider: "{provider_name}"
      language: "{language}"
      generated: "{date}"
      arc_type: "{arc_type}"
      arc_id: "{arc_id}"
      governing_thought: "{governing_thought}"
      confidence_score: {avg_confidence}
      transformation_notes: |
        Story-to-slides transformation.
        Theme: {theme_id}. Arc: {arc_type}.
        {N} slides, {avg}% avg confidence.
        {number_plays} number plays, {headlines_optimized} headlines optimized.

    TITLE: {title}
    SUBTITLE: {subtitle}

    SLIDE_SPECS:
    {all slide YAML from Steps 8 + 8.1}

    AUDIENCE_MODEL:
    {audience model from Step 3}

    ARC_ANALYSIS:
    {arc analysis from Step 4}

    LANGUAGE: {language}
    ARC_ID: {arc_id or "none"}
    ARC_DEFINITION_PATH: {path or "none"}
    BUYER_APPENDIX_PATH: {path or "none"}

    CTA_SUMMARY:
    {cta_summary from Step 6.1 or "none"}

    GENERATION_METADATA_STATS:
      number_plays: {count}
      headlines_optimized: {count}
      bullets_consolidated: {count}
      source_links: {count}
      layout_distribution: "{layout_type: count, ...}"
      avg_confidence: {score}
      manual_review: [{slide list or "none"}]

On success (ok: true): The agent wrote the complete brief to output_path. Read back the first 30 lines to confirm the file exists and has correct frontmatter. Skip Step 10 — the brief is already written. Proceed directly to Step 9 (validation).

On failure (ok: false): Log the error. Fall back to inline execution: read references/08c-presenter-prep.md and references/05b-speaker-notes.md, generate prep slides and speaker notes inline, insert prep slides after Slide 1, renumber, append speaker notes. Then proceed to Step 10 to write the brief.

---

Step 9: Validate Against Schema

Self-assessment is unreliable without explicit measurement — in early tests, models reported "pass" while producing topic-label headlines and missing citations. The five-layer gate forces honest evaluation.

Read reference: references/09-validation-checklist.md

Five layers — stop on first failure, fix, re-check:

1. Schema — layout types exist, required fields present, no visual fields, valid YAML
2. Message quality — assertion headlines, MECE sequence, isolated hero numbers
3. Copywriting — number plays applied, bullets consolidated, no hedging
4. Presentation logic — bookend slides enforced, within max_slides, layout variety
5. Content integrity — all sections represented, citations preserved, German characters correct

---

Step 9b: Stakeholder Review (when stakeholder_review=true)

Structural validation (Step 9) catches schema and formatting issues, but cannot tell whether the brief will actually work for the audience, the presenter, or as a visual communication. The brief-review-assessor evaluates from three stakeholder perspectives — catching weak headlines that pass schema checks, layout monotony that passes variety rules, and CTA gaps that pass structural validation. Reviewing at the brief stage is efficient because changes are text edits, not re-renders.

Skip this step if stakeholder_review=false.

Launch the brief-review-assessor agent with:

• brief_type: slides
• Brief content at output_path (the file was written by the enrichment agent in Step 8.2, or will be written in Step 10 on the fallback path)
• source_narrative: the narrative path from Step 0
• audience_context: if provided
• round: 1

On accept (all perspectives ≥85): Proceed to Step 10 (or Step 11 if brief already written).

On revise:

1. Apply CRITICAL improvements first, then HIGH improvements — edit the brief content surgically (change specific headlines, layout types, speaker notes, CTAs as recommended)
2. Re-run Step 9 validation to ensure structural integrity after edits
3. Re-launch the assessor (round 2)
4. If round 2 accepts or scores 70+ with no CRITICAL issues: proceed to Step 10
5. If round 2 still has issues: present remaining issues to user, proceed to Step 10

On reject: Surface the verdict to the user via AskUserQuestion and let them decide whether to proceed, edit manually, or abandon.

Write the review verdict to {output_dir}/presentation-brief.review.json.

---

Step 10: Write Presentation Brief (fallback path only)

If Step 8.2 succeeded (ok: true), the enrichment agent already wrote the complete brief. Skip this step and proceed to Step 11. This step only executes on the fallback path (agent failure → inline enrichment).

Read reference: references/07-output-template.md (Brief Output Template section)

Output path resolution (run via Bash before writing):

• If output_path explicit: mkdir -p "$(dirname "${output_path}")"
• Otherwise: set output_path = {source_dir}/cogni-visual/presentation-brief.md and mkdir -p "{source_dir}/cogni-visual"

Write the brief following the output template. YAML frontmatter must include: type, version (4.0), theme, theme_path, arc_type, arc_id, governing_thought, confidence_score. Include PPTX Rendering Requirements section (localized). Include Generation Metadata at end.

Run the validation checklist (reference 09-validation-checklist.md) one final time against the written file. The most commonly missed items at this stage: mispositioned PPTX Rendering Requirements section, and superscript <sup>[N](url)</sup> in body text vs plain [N](url) in speaker notes.

---

Step 11: Guide User to PPTX Rendering

The presentation brief is ready. Now guide the user to the best rendering path — currently claude.ai chat with the Anthropic PPTX skill, which produces the highest quality output.

After the brief is written and validated, tell the user:

1. Open a new chat on claude.ai (not Claude Code — the PPTX skill works best in the claude.ai web interface)
2. Paste these two files into the chat window as attachments:
   • presentation-brief.md (the brief you just generated)
   • theme.md (the theme file used in this workflow)
3. Use this prompt:

Please create a PPTX presentation using the attached presentation-brief.md and theme.md

Print the absolute paths to both files so the user can locate them easily:

─── Files to attach in claude.ai ───

Presentation brief: {absolute_path_to_presentation_brief}
Theme:              {absolute_path_to_theme_md}

Open claude.ai → new chat → attach both files → paste the prompt above.
─────────────────────────────────────────────────

Replace {absolute_path_to_presentation_brief} with the resolved output_path and {absolute_path_to_theme_md} with the theme_path from Step 1. Both paths must be absolute — never use ~, $HOME, $CLAUDE_PLUGIN_ROOT, or relative paths.

Why claude.ai? The Anthropic PPTX skill in the claude.ai web interface handles file attachments natively and produces the best slide rendering results. Claude Code can also render via the document-skills:pptx skill, but the claude.ai path is currently the recommended workflow for highest quality output.

---

Bundled Resources

References (loaded at specific steps — progressive disclosure)

Reference	Step	Purpose
02-audience-model.md	3	Audience Model construction (Rich/Lean mode)
03-story-arc-analysis.md	4	Arc detection, governing thought, section roles
04-message-architecture.md	5	Pyramid Principle, one-message-per-slide, MECE, consolidation
05a-slide-copywriting.md	6	Assertion headlines, number plays, bullet consolidation
05b-speaker-notes.md	8.2	Two-section speaker notes format reference (loaded by slides-enrichment-artist agent)
06-slide-mapping-rules.md	7	Layout selection, confidence scoring, fallback strategies
07-output-template.md	8, 8.2, 10	Slide YAML example, brief output template, citation rules (also loaded by slides-enrichment-artist agent)
08b-references-slide.md	8.1	References slide construction
08c-presenter-prep.md	8.2	Internal prep slides + per-slide speaker notes process (loaded by slides-enrichment-artist agent)
09-validation-checklist.md	9	Five-layer validation framework
2g-diagram-simplification.md	2.1	Mermaid diagram detection and simplification

Libraries (loaded as needed — progressive disclosure)

Library	Step	Purpose
arc-taxonomy.md	1	Arc ID → visual arc type mapping, element names
cta-taxonomy.md	1	CTA types, urgency, arc-to-CTA heuristics
pptx-layouts.md	7	Slide layout schemas and field definitions (deferred from Step 1)
EXAMPLE_BRIEF.md	8	Output format reference (deferred from Step 1)

Backward Compatibility

• arc_type: why-change activates Why Change file discovery
• project_path parameter still accepted (mapped to source_path)
• Power Position extraction still supported