SkillNewofficial publisher

Image to Editable PPT (Ming-Image)

Skill : A reusable method: instructions, procedures, sometimes scripts or templates

Agent skill that recreates an AI-generated slide image as an editable PowerPoint presentation, with texts and simple shapes converted to native elements.

Useful for escaping the trap of an AI-generated slide that can no longer be edited: the agent reconstructs the page in PowerPoint. A compatible model service is required (self-hosted Ming-Image or via an API): where your data goes depends on that choice.

Official site GitHub · 212 ★ Docs

Overall score

6.8 /10

Security: one caveat, see the card

Utility

7.0

Reliability

10.0

Security

One caveat

Ease

Technical · Technical

Security reflects where data goes, requested permissions and incidents on record at the verification date. It is not a guarantee of zero risk: check the deployment mode and the connections you enable.

What it does, concretely

When an AI agent generates a slide as an image, the result is locked: it is impossible to correct a text, move an element, or change a color without starting over. Image to Editable PPT solves this by reconstructing the slide in a native PowerPoint file, where every text and simple shape becomes an editable element again.

The skill integrates into a Ling agent and relies on the Ming-Image-0.1-Design-Layer model to decompose the image into distinct transparent layers. These layers are then translated into native PowerPoint elements (text boxes, shapes) and assembled into a ready-to-use .pptx file.

This workflow is aimed at people who produce presentations with generative AI tools and need to retouch the result in PowerPoint without starting from scratch. It also suits teams that want to automate slide production while keeping control over the final content.

It is not suited for those looking for a turnkey online presentation editor: this is a technical skill that requires setting up a Python environment, a Ling agent, and a compatible model service (local or via API).

What it offers

·Converts a slide image into a PowerPoint file (.pptx) with native editable elements.
·Decomposes the image into independent transparent layers using the Ming-Image-0.1-Design-Layer model.
·Converts detected text into editable PowerPoint text boxes.
·Converts simple shapes into native PowerPoint shape elements.
·Integrates as a skill into a Ling agent, usable in an automated workflow.
·Compatible with local deployment (own GPU) or via an external API service (OpenRouter at launch).
·Open-source code under MIT licence, freely reusable and modifiable.

Our tips

For layer decomposition, using a resolution of 1024 gives better results; resolution 512 is faster but less precise.
The layer decomposition model requires a GPU with at least 80 GiB of memory to run under the conditions validated by the authors.
If you use an external model service (such as OpenRouter), check that the image sent does not contain sensitive data, as it passes through that third-party service.
The skill follows its own setup instructions in its repository (ling-cookbook): consult the README of the image-to-editable-ppt skill for exact dependencies, which are separate from those of the main Ming-Image model.
FlashAttention 2 is optional but speeds up inference if your environment supports it; the CLI defaults to eager mode, which is more portable.

Watch out

The minimum validated deployment requires a GPU with at least 80 GiB of GPU memory, which rules out most consumer-grade setups.
Complex shapes or highly elaborate layouts may not be faithfully reconstructed as native PowerPoint elements, as conversion is limited to text and simple shapes.

In practice

Price: Free
Commercial use: Commercial use: yes
Available on: GitHub
Licence: MIT
FR interface: No

Security and data

Rating: Security: one caveat, see the card
Your data: Local or cloud depending on use
Maintenance: Active

Links and repository checked on 18 October 2026 · added on 29 September 2026

How to install Image to Editable PPT (Ming-Image)

1. Clone the ling-cookbook repository and navigate to the image-to-editable-ppt skill folder. 2. Install Python 3.10 or newer in a dedicated virtual environment. 3. Install dependencies with: pip install -r requirements.txt 4. Configure the model service (local deployment of Ming-Image-0.1-Design-Layer on GPU, or an OpenRouter API key). 5. Follow the Ling agent setup instructions described in the skill's README. 6. Provide a slide image to the agent: the editable .pptx file is generated as output.

pip install -r requirements.txt

Official documentation ↗

In the same category

Whole category Design & slides →