Paperbanana Slide Deck logo

Paperbanana Slide Deck

Community
PlutoLei
paperbanana-slide-deck

Create or revise a complete PaperBanana slide deck, including image slides and native editable PowerPoint text, tables, charts and diagrams. Use for PaperBanana deck requests; preserve the user's selected presentation tool and existing deck mode.

Overview

PublisherPlutoLei
Repositorypaperbanana-skill
Skill namepaperbanana-slide-deck
Stars
60
Forks
3
Bundled files
142
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 142 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by PlutoLei on GitHub. Read the source before you install it.

Installation

Install the Paperbanana Slide Deck AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/PlutoLei/paperbanana-skill/tree/master/plugins/paperbanana-slide-deck
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/PlutoLei/paperbanana-skill.git /tmp/paperbanana-skill
mkdir -p .claude/skills
cp -r /tmp/paperbanana-skill/plugins/paperbanana-slide-deck .claude/skills/paperbanana-slide-deck
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Paperbanana Slide Deck in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Paperbanana Slide Deck on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Paperbanana Slide Deck is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

PaperBanana slide deck

Two output modes share content planning and final inspection:

  • image: per-slide prompts → generated full-slide images → PPTX assembly.
  • editable: slide-spec.json → native PowerPoint objects → rendered previews → review → targeted spec edits → rebuild.

Resolve this bundle

BUNDLE_DIR is the absolute directory containing this file, package.json, scripts/ and references/. A plugin host may discover the small entrypoint in skills/paperbanana-slide-deck/; that entrypoint points here, two levels above it. Use the path supplied by the host, not a search of another runtime's directories.

Before building, verify scripts/build-deck.ts exists and run bun install --frozen-lockfile in BUNDLE_DIR when dependencies are absent. Only copying the nested discovery file is not a complete installation. The local editable renderer needs Bun, not an image API or Python backend.

Select mode and scope

For a plan or read-only review, return the requested analysis without writing a deck or calling generation. Otherwise reuse the chosen mode:

RequestMode
Edit native text, cells, charts or connectors in PowerPointeditable
Generate full-slide artwork or continue an existing image deckimage
A new deck with no output-mode preferenceExplain the difference and resolve that choice before generation

Record the chosen mode in the task's existing notes or deck-mode.txt. Do not silently fall back from editable to image mode. Inspect an existing deck and its inputs before changing it; missing outline.md alone does not mean it is new.

For a new deck, establish audience, source material, slide count and style. Reuse approved choices and outline. Read the outline contract and style guide as needed. Resolve material missing choices, then author the outline and prompts/spec. A scoped revision does not need a new full-deck outline or style-selection interview.

Editable workflow

Read the spec contract before authoring.

  1. Author slide-spec.json with native text, shapes, lines, tables and charts. Use image assets only where the task needs them; diagrams do not require a preliminary image-model call. Preserve research facts and label example data. Include sources in speaker notes and the schema's required source fields.

  2. Validate and build from any working directory:

    bash
    bun "$BUNDLE_DIR/scripts/build-deck.ts" --mode editable "$DECK_DIR"
  3. Render the current spec through the bundled preview helper, which rebuilds the PPTX and associates spec, slides and preview files in a manifest:

    bash
    python3 "$BUNDLE_DIR/scripts/editable/render-previews.py" --spec "$DECK_DIR/slide-spec.json" --output-dir "$DECK_DIR/rendered"

    Preview rendering needs LibreOffice and Poppler. If unavailable, deliver the built PPTX with its visual inspection explicitly pending; do not invent previews.

  4. Inspect every preview for text fit, cropping, object placement and content. Automated Critic review is optional when not requested and requires a configured PaperBanana Python backend plus authorization for its VLM calls:

    bash
    python3 "$BUNDLE_DIR/scripts/editable/review-editable.py" --spec "$DECK_DIR/slide-spec.json" --rendered-dir "$DECK_DIR/rendered" --output "$DECK_DIR/critic-review.json"
  5. Preserve unresolved/unmapped findings. Edit the affected elements in the spec, rebuild and rerender before reviewing again. Never use old previews or a previous review as evidence for a changed deck. Do not patch the generated PPTX as the sole source of a fix while leaving the spec stale.

Invalid specs, missing assets and hash mismatches must fail without an image-mode fallback. Native OOXML objects prove structure; rendered previews prove visible layout; actual PowerPoint interaction is needed to claim that application was tested.

Image workflow

  1. Keep the approved style tokens across prompts/NN-slide-slug.md: palette, type hierarchy, background and recurring visual treatment. Keep required labels, quantities and connections explicit.

  2. Use the configured PaperBanana backend and preserve explicit/saved provider and model choices. Check slide --help or slide-batch --help for supported flags. Credentials stay in the configured environment, outside prompts/logs.

    bash
    python -m paperbanana.cli slide-batch --prompts-dir "$DECK_DIR/prompts" --output-dir "$DECK_DIR"
  3. For partial batches, preserve completed pages and resume only missing pages. For one-page edits, regenerate only that page with the existing style tokens. Do not assume slide-batch --continue is supported. If the backend cannot target a subset, stage the requested prompt(s) in a separate directory and reconcile only those outputs; don't overwrite unrelated slides.

  4. Before assembly, compare the intended slide list with actual image filenames. Report missing, duplicate or orphaned pages. Do not silently delete files or merge a stale orphan; stage the intended images/prompts for assembly when needed.

  5. Assemble and inspect the result:

    bash
    bun "$BUNDLE_DIR/scripts/build-deck.ts" --mode image "$DECK_DIR"

Critic failure does not require generating the image again. Preserve it as UNREVIEWED. Do not switch to a direct Gemini script or another provider solely because the selected route failed. Reuse approved alternatives when present; otherwise report the limitation and resolve the missing choice.

Delivery and recovery

Return the deck, inspected previews and the status of each requested output. List missing pages and review gaps. A model score is not scientific acceptance. Do not claim native editing when the deck contains full-slide images.

Retry recoverable failures within scope after diagnosing them. Respect the approved call limits and provider choices; do not pursue an unbounded "until clean" loop. Further revisions should address concrete findings. Preserve source specs, prompts and existing user edits throughout.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

and 82 more files.

Frequently asked questions

What does the Paperbanana Slide Deck AI skill do?

Create or revise a complete PaperBanana slide deck, including image slides and native editable PowerPoint text, tables, charts and diagrams. Use for PaperBanana deck requests; preserve the user's selected presentation tool and existing deck mode.

Why use Paperbanana Slide Deck on TypingMind?

Because you install it once and use it with any model. Paperbanana Slide Deck is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Paperbanana Slide Deck in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/PlutoLei/paperbanana-skill/tree/master/plugins/paperbanana-slide-deck. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Paperbanana Slide Deck?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Paperbanana Slide Deck?

As many as you like. As long as a model supports skills, you can use Paperbanana Slide Deck with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Paperbanana Slide Deck AI skill free?

Yes. It is published on GitHub by PlutoLei under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇