Guide · Agent skill
How to use the Scientific Slides Builder agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The scientific-slides skill teaches an AI agent how to build slide decks for research talks, including conference presentations, seminars, thesis defenses and journal clubs. It covers slide structure, design, timing and visual checks, and it can output editable PowerPoint or LaTeX Beamer files. This guide explains what the skill does, how to install it, how to ask for help in natural language, and what to check before you present.
Quick start
- 1Download and unzip. Download the skill ZIP from the directory page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder.
- 2Or upload in the Claude apps. In the Claude apps, open Settings, then Capabilities, then Skills, and upload the skill.
- 3Install the required tools. Install the Python packages the skill lists, plus TeX Live or MiKTeX, Node.js with PptxGenJS, or LibreOffice, depending on the outputs you need.
- 4Set the API key if you need AI images. Set OPENROUTER_API_KEY only if you plan to generate image slides. It is optional for the other workflows.
- 5Describe your talk. Tell the agent the talk type, length, audience and output format, and attach your figures and data.
- 6Plan structure and timing. Review the slide plan and timing before the agent designs the slides, so the deck fits your allotted time.
- 7Review the rendered slides. Ask the agent to run its visual review, then check every number, label and citation yourself before presenting.
What an agent skill is and what this one changes
An agent skill is a folder of instructions and supporting files that an AI agent loads when a request matches the skill's description. This skill's description covers PowerPoint slides, conference presentations, seminar talks, research presentations, thesis defense slides and any scientific talk.
Once the skill is active, the agent follows its guidance: it plans slide structure and timing, applies scientific design principles, and can produce slides as editable PowerPoint or LaTeX Beamer files. It can also generate AI image slides and combine them into a PDF, and it includes checks for reviewing rendered slides for layout and consistency.
- Slide structure and narrative planning
- Timing guidance, including a one-slide-per-minute rule of thumb
- Design templates and figure guidance for data visualization slides
- PowerPoint (via PptxGenJS) and LaTeX Beamer output
- Visual validation of rendered slides
Install the scientific-slides skill
Option one: download the skill ZIP from the directory page and unzip it into your personal skills folder, ~/.claude/skills/. To make it available only inside one project, unzip it into that project's .claude/skills/ folder instead.
Option two: in the Claude apps, open Settings, go to Capabilities, and upload the skill under Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills
Install the software the skill needs
The skill runs Python scripts and needs several Python packages: requests, Pillow, PyMuPDF, pypdf and python-pptx. Check the skill's requirements for the exact Python version it expects.
Some outputs need extra tools. Beamer needs TeX Live or MiKTeX. Programmatic PowerPoint needs Node.js and PptxGenJS. Rendering PowerPoint files for review needs LibreOffice. Image slide generation needs an OpenRouter API key in the OPENROUTER_API_KEY environment variable and a network connection; this key is optional and only needed for that step.
- Required: Python packages requests, Pillow, PyMuPDF, pypdf, python-pptx
- Beamer output: TeX Live or MiKTeX
- Programmatic PPTX: Node.js and PptxGenJS
- Review rendering: LibreOffice
- Optional: OPENROUTER_API_KEY for AI image generation
How to trigger the skill with a natural request
You do not need special syntax. Describe the talk in plain language, say what kind of talk it is, and say what format you want. Give the agent your own figures and data where you have them, since the skill's guidance says to locate the user's figures and underlying data before building results slides.
Include the presenter name and affiliation if you want them on the title slide. The skill says to omit them if they are unknown.
- 'Make a 20-minute conference deck in PowerPoint on my protein folding results.'
- 'Build a Beamer seminar talk for 45 minutes from these three figures.'
- 'Plan the structure and timing for my thesis defense before we design any slides.'
- 'Create a journal club deck on this paper with a dark blue, minimal style.'
Example: generate a title slide image from the command line
For AI image slides, the skill uses the generate_slide_image.py script. The skill's own example states the formatting goal in the prompt and saves the output to a file. Use --attach with the previous slide so later slides match the style of earlier ones.
Image slides are drafts. The skill warns that generated images may redraw data, labels or citations incorrectly, so verify every claim before presenting.
python scripts/generate_slide_image.py "Title slide for presentation: 'Machine Learning: From Theory to Practice'. FORMATTING GOAL: Dark blue background, white text, gold accents, minimal design." -o slides/01_title.pngChoose the right output for your slides
Use the image workflow for conceptual or overview slides. Use editable PowerPoint or Beamer for quantitative results, exact equations, accessible text or a template you supply. The skill recommends embedding the original chart unchanged, or regenerating it from source data, rather than relying on an image model to redraw a figure.
The skill's guidance on citations says to include them in the slide prompt in small text, such as (Author et al., Year), and to list multiple citations together.
- Conceptual or overview slides: AI image slides combined into a PDF
- Results, equations, accessibility or templates: editable PowerPoint or Beamer
- Final results: original figure plus verified text
Safety notes: scripts and network use
This skill runs scripts and uses the network. Image generation sends prompts to OpenRouter and needs OPENROUTER_API_KEY. Read the scripts and the skill's files before running them on material you cannot share with an outside service.
Keep unreviewed output out of your talk. The skill's own warning is that a generated image can alter numbers, labels, citations and scientific relationships.
- Generation requires network access and your own API key
- Do not use the image workflow for final numerical results
- Check every citation against its source
What people use it for
- A graduate student building a thesis defense deck from their own figures and a planned narrative.
- A researcher preparing a 45-60 minute seminar talk, with timing guidance for each section.
- A presenter creating a short conference talk of about 5-20 minutes.
- A journal club leader planning slides for a paper, starting with the talk's structure.
- A team converting a result set into an editable Beamer deck for a lab meeting.
Frequently asked questions
Is the scientific-slides skill free?
The skill is published under the MIT license, as stated on its page. Some optional steps use OpenRouter, which requires an API key and network access for image generation.
What do I need installed to use it?
You need Python with requests, Pillow, PyMuPDF, pypdf and python-pptx. Beamer needs TeX Live or MiKTeX, programmatic PowerPoint needs Node.js and PptxGenJS, and rendering needs LibreOffice.
Where do I put the skill folder?
Unzip the downloaded ZIP into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder for one project. In the Claude apps, you can upload it under Settings, Capabilities, Skills.
Do I need an OpenRouter API key?
Only for AI image generation. The source marks OPENROUTER_API_KEY as optional and says it is needed only for the skill's LLM-powered image steps, which also need network access.
Can it make PowerPoint and LaTeX Beamer slides?
Yes. The skill works with PowerPoint and LaTeX Beamer, and it also supports AI-generated image slides combined into a PDF. Editable PowerPoint or Beamer is the recommended route for exact equations and quantitative results.
Can I trust the slides for final results?
Review everything. The skill warns that the image workflow is a draft generator that may redraw data, labels or citations incorrectly, so use the original figures and verified text for results.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits graduate students and researchers who want help planning a talk's structure and timing, and then building editable PowerPoint or Beamer slides from their own figures. It is a poor fit if you need the image workflow for final numbers, or if you cannot verify every claim and citation yourself or lack the listed software.
Download the skill