Skip to content
Skills Library
Documents & office · Conversion & formats

extract

/extract

Extracts a team's design system from an original PowerPoint deck into a shareable pptx-kit design pack (source intake, archetype pick, parallel per-slide rebuild verified against PowerPoint renders, then consolidation into DESIGN.md, catalog.json, tokens.json and shared components). Long-running and usage-heavy; run only when the user asks to extract or build a pack.

Repository
pptx-kit ↗
Licence
MIT
Last repository change
09.11.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 09.11.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Extracts a team's design system from an original PowerPoint deck into a shareable pptx-kit design pack (source intake, archetype pick, parallel per-slide rebuild verified against PowerPoint renders, then consolidation into DESIGN.md, catalog.json, tokens.json and shared components). Long-running and usage-heavy; run only when the user asks to extract or build a pack.

Arguments

/extract <source.pptx> <packDir> [--yes] [--max N]
Original author description

Extracts a team's design system from an original PowerPoint deck into a shareable pptx-kit design pack (source intake, archetype pick, parallel per-slide rebuild verified against PowerPoint renders, then consolidation into DESIGN.md, catalog.json, tokens.json and shared components). Long-running and usage-heavy; run only when the user asks to extract or build a pack.

How it works

  1. 1. Preflight
  2. 2. Intake
  3. 3. Archetype pick
  4. 4. Rebuild (subagents)
  5. 5. Verify
  6. 6. Consolidate
  7. 7. Finish
Original SKILL.md headings · 57 lines, 756 words
SKILL.md preview
---
name: extract
description: Extracts a team's design system from an original PowerPoint deck into a shareable pptx-kit design pack (source intake, archetype pick, parallel per-slide rebuild verified against PowerPoint renders, then consolidation into DESIGN.md, catalog.json, tokens.json and shared components). Long-running and usage-heavy; run only when the user asks to extract or build a pack.
argument-hint: "<source.pptx> <packDir> [--yes] [--max N]"
disable-model-invocation: true
---

# Extract a design pack from a source deck

Arguments: `$ARGUMENTS` → `<source.pptx> <packDir>`; `--yes` skips the archetype confirmation; `--max N` caps archetypes (default 25).

Plugin root: `${CLAUDE_PLUGIN_ROOT}`. Plugin files you open with Read (another SKILL.md, agents/*.md) still show the unresolved plugin-root placeholder; it stands for that same path. Read `${CLAUDE_PLUGIN_ROOT}/reference/pack-format.md` first (pack layout, CLIs). Everything runs locally: PowerPoint, Node, Python and this Claude Code session; nothing is uploaded anywhere.

Ask for the two paths if missing. Keep `<packDir>` on an ASCII path when possible. Use absolute paths in every command and in every task you give a subagent.

## 1. Preflight

`node "${CLAUDE_PLUGIN_ROOT}/tools/doctor.mjs"` — stop on any failed check except `ascii working dir` and explain the fix (see the setup skill: `/pptx-kit:setup`).

## 2. Intake

`node "${CLAUDE_PLUGIN_ROOT}/tools/intake.mjs" --src "<source.pptx>" --pack "<packDir>"`

- Missing fonts → stop. Tell the user which faces are missing; they install them (per-user install is enough) and you rerun. Only pass `--allow-missing-fonts` if the user explicitly accepts substitute-font renders (the pack will then be measured against wrong renders).
- Output: `facts/sNN.json`, `facts/media/`, `facts/deck-stats.json`, `src/png/sNN.png`, `src/sheet-N.png`, `pack.json`.

## 3. Archetype pick

1. Read `facts/deck-stats.json` and view every contact sheet `src/sheet-*.png` (full deck).
2. Classify every slide by visual structure and slide kind: cover, contents, section divider, subsection / interstitial, content variants (by body pattern: quote cards, table, screenshot + numbered findings, comparison columns, flow, matrix, image grid, charts drawn from shapes …), insert / example slides, closing. View a slide at full size (`src/png/sNN.png`) when a thumbnail is ambiguous.
3. Pick ONE representative per distinct structure — the instance with the most content (it defines capacity), clean of one-off decoration. Cover every slide kind the deck uses. Typical result: 10–25 archetypes; never more than `--max` (default 25) without the user's consent.
4. Pick the **seed**: a typical content slide with the full chrome (breadcrumb / header, tracker, page number). The seed's rebuilder owns `tokens.json` and `lib/` (chrome + shared helpers); all others only read them.
5. Write `<packDir>/_work/archetypes.json`:
   `{ "seed": 11, "archetypes": [ { "slide": 11, "id": "s11-quote-columns", "kind": "content", "why": "...", "alsoLike": [12, 14, 40] } ] }` (`id` = `sNN-<short-purpose>`, NN zero-padded to 2 as in the PNG names).
6. Unless `--yes`: show the table (slide, id, kind, why, how many slides it stands for) and the estimate — each archetype is one long subagent run (roughly 20–60 min wall time in parallel batches, and a large share of a subscription's usage window for a full deck). Ask the user to confirm or edit the list. Do not start rebuilding before the answer.

## 4. Rebuild (subagents)

Each archetype is rebuilt by the `pptx-kit:slide-rebuilder` agent. Task text per agent:

> Pack: `<abs packDir>`. Rebuild source slide sNN as layout `<id>` (kind: <kind>; similar slides: <alsoLike>). You [own | do not own] `tokens.json` and `lib/`. Follow your method; finish with the report it asks for.

1. Seed first, alone. When it finishes, check `tokens.json` and `lib/*.mjs` exist and the seed renders (`render.mjs --compare`).
2. Then all other archetypes, at most 4 agents running at once (PowerPoint renders queue on one lock; more agents only wait). Launch the next as one finishes.
3. If the `pptx-kit:slide-rebuilder` agent type is unavailable (pptx-kit opened as a plain folder), use a general-purpose agent and prefix the task with: "Read `${CLAUDE_PLUGIN_ROOT}/agents/slide-rebuilder.md` and follow it as your instructions; its unresolved plugin-root placeholder stands for `${CLAUDE_PLUGIN_ROOT}`."

## 5. Verify

For every layout, sequentially: `node "${CLAUDE_PLUGIN_ROOT}/tools/render.mjs" --pack "<packDir>" --layout <id> --out "<packDir>/_work/verify/<id>" --compare "<packDir>/src/png/sNN.png"`. Also render `stress-max.json` and `stress-min.json` with `--data` (overflow only). Build a table: id, ssim, overflow (sample / max / min), errors.

- Fails = errors, any overflow, or ssim < 0.95 → send that layout back to a rebuilder once, with the failing JSON lines: "Continue from the existing files in `layouts/<id>/`; fix: …". Re-verify. Report what still fails; do not loop further.
- View 3–4 `compare.png` files yourself (lowest ssim first) — a high score can hide a wrong line break.

## 6. Consolidate

Read `${CLAUDE_PLUGIN_ROOT}/skills/consolidate/SKILL.md` and follow it for `<packDir>` (it writes tokens.json final, lib/components.mjs, lib/demo.mjs, catalog.json, DESIGN.md, pack.json fonts).

## 7. Finish

1. `node "${CLAUDE_PLUGIN_ROOT}/tools/pack.mjs" prune "<packDir>"` (dry run) — show removed MB; then `node "${CLAUDE_PLUGIN_ROOT}/tools/pack.mjs" prune "<packDir>" --apply`.
2. `node "${CLAUDE_PLUGIN_ROOT}/tools/pack.mjs" check "<packDir>"` — fix every error.
3. Report: archetypes and their final ssim / overflow, components added, unresolved items, pack size, fonts the receiver must install (with sources). Tell the user the pack folder is the shareable unit (copy the folder) and that the next step is `/pptx-kit:generate <packDir> <contentDir>`.

Versions

RepositoryScoreLicence
This page · sttsspjy/pptx-kit100MIT
alirezarezvani/claude-skills100MIT
actionbook/actionbook100Apache-2.0
filiphsps/commerce100MIT
akarachen/aghub100MIT
athola/claude-night-market100MIT
aladicf/better-web-ui100MIT
adobe/skills95Apache-2.0
educlopez/ui-craft92MIT

Database snapshot: 13.11.2026 · package version 1398c2424a03 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 95pptx-skill
    Create, edit, and extract content from PowerPoint (.pptx) files; use when you need to generate slides programmatically, update existing decks, or export slide previews.
  • 90Extract PPTX Content and Structure
    Use this skill to extract text, slide titles, and content from PowerPoint presentations to accurately determine their subject matter for classification.
  • 100powerpoint
    Creates, edits, and reads PowerPoint (.pptx) files programmatically with the python-pptx library. Use when a user asks to generate a PPTX, modify slides, extract text from a presentation, add charts or tables to PowerPoint, build slides from data, convert content to PPTX, or…
  • 100powerpoint
    Create professional PowerPoint presentations. Use when asked to "create a PowerPoint", "make a presentation", "build slides", or generate pptx files.
  • 100deck-pptx
    Build an editable PowerPoint (.pptx) presentation from a topic or an outline. Use it when the user asks for a deck, slides or a presentation and wants a .pptx file they can open and edit.
  • 90Presentations
    Read, create or edit PowerPoint or Google Slides decks. Use for presentation, slide deck, PowerPoint, PPT, PPTX, or Google Slides requests.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.