Jiutian Knowledge Graph and Knowledge Cards

Problem It Solves

When turning a topic, article, or notes into a knowledge graph, the hard part is often not a lack of information, but making scattered concepts explicit, modeling their relationships, and producing a shareable artifact. This skill targets that workflow: given a concept or content, it chooses between card and graph output, generates a browser-renderable HTML page, and supports screenshot sharing or PNG export.

Core Capabilities and Workflow

• Template selection: It includes 17 layouts, such as classic cards, checklists, timelines, comparison tables, radial graphs, hierarchy trees, flow chains, mind maps, and metro maps, covering concept summaries, method lists, comparisons, process relationships, and complex knowledge networks.
• Size adaptation: Before generation, it confirms mobile one-screen or desktop default mode. The first is compact and suited to mobile sharing; the second keeps normal spacing for desktop viewing or printing.
• Information enrichment: The skill can use WebSearch, knowledge-base retrieval, and internal knowledge to organize definitions, key points, underlying logic, application scenarios, and related concepts, with cross-validation when sources allow.
• Relationship modeling: For graph templates, it extracts core concepts, sub-concepts, relationship labels, and node connections into a renderable SVG structure, useful for showing contains, depends on, causes, or process relationships.

Boundaries and Caveats

It works best when the topic is clear enough to structure. Broad inputs may require a confirmation step to narrow the focus and template. Long inputs are usually summarized before modeling. If retrieval tools are unavailable, it may fall back to general knowledge and mark the output as not real-time validated. If relationships are weak or cannot be extracted reliably, the result may degrade to a card template.