This repository is primarily an agent-skill repository, not a generic command-line graphics toolkit.
The main deliverable is reusable skill packaging under skills/, so agents can accept natural-language requests such as:
- "Turn this text into an editorial info card."
- "Make this into a 4:3 information card."
- "Use the editorial-card-screenshot skill and output HTML plus a PNG."
For the current repository, the installable unit is the full skill directory:
skills/editorial-card-screenshot
Agents should install or copy the whole package rather than a single file.
The package includes:
SKILL.mdassets/references/scripts/agents/when supported by the host agent
The shell script and references are support files for the skill. They are not meant to be installed independently.
For the current editorial-card-screenshot skill:
- the user gives source material in natural language
- the agent follows the behavior contract in
SKILL.md - the agent applies the stored editorial prompt rules
- the agent generates the HTML card
- if image output is requested, the agent uses the bundled screenshot helper internally
skills/editorial-card-screenshot/scripts/capture_card.sh exists to support rendering, but it is not the primary user interface.
The primary interface is the agent itself.
After installation, a good verification prompt is:
Use $editorial-card-screenshot to make a 3:4 editorial info card from these notes.
Other useful prompts:
Use $editorial-card-screenshot and give me HTML only.Use $editorial-card-screenshot and render both HTML and PNG.
Today the repository contains one packaged skill:
editorial-card-screenshot
Over time, more information-card skills can be added under skills/ using the same structure.