For AI agents
The Fluxygram skill teaches an AI agent the format, how to choose the diagram type and how to write steps that explain. With it, you can say “make an animated diagram of how OAuth works” and get a valid .flux.json.
The repository is a plugin marketplace:
/plugin marketplace add Fluxygram/fluxygram/plugin install fluxygram@fluxygramThen just ask: “Make an animated diagram of how our checkout works.”
Save the skill in your project and point the agent at it (or add it to your rules):
curl -o FLUXYGRAM.md https://fluxygram.app/skill.mdThen ask: “Following FLUXYGRAM.md, diagram how a request goes through our API.”
On the Fluxygram site, pick your assistant, type what you want to explain and press Copy prompt. The prompt includes the whole skill. Paste it in the chat, then paste the JSON answer into the editor.
What the agent learns
Section titled “What the agent learns”- Which of the seven types fits the question.
- The format: blocks, connections, groups, icons, states.
- The storytelling rules: one idea per step, short captions,
flowfor real movement,statefor outcomes. - A recipe to diagram a whole codebase.
The full skill is at skills/fluxygram/SKILL.md, and also served at fluxygram.app/skill.md.
These docs, for agents
Section titled “These docs, for agents”These docs are also published as plain text for language models:
/llms.txt: a short overview that links to the two versions below./llms-full.txt: all the documentation in one file./llms-small.txt: a shorter version, for small context windows.