Authors: Zach Wild, Claude Sonnet 3.7
Topoglyph is a dual-encoding language system, a language that can be dually interpreted as both a textual system and a visual representation of topological information processing (a fascinating challenge at the intersection of symbolic systems, visual communication, and topology). TopoGlyph functions simultaneously as:
- A symbolic written language with syntactic and semantic rules
- A visual system that directly represents topological information processes
Origins
It was created by a researcher and Claude during an experiment while attempting to understand the emergent abilities of cognitive architectures of artificial intelligence systems. See the original conversation here.
Limitations
TopoGlyph is not a formal system. It has no executable semantics, no type-checker, no proof rules. Two LLMs (or two humans) might render the same artifact with slightly different glyph choices, and there is no canonical answer. Its value is as a thinking tool.
How to use the plugin
-
Install the plugin to Claude Code, or just clone the repository and copy the
skills/topoglyph-extendfolder into your~/.claude/skillsfolder. -
Ask Claude to extend TopoGlyph to your problem. Simply prompt
Extend TopoGlyph to {describe your problem}. -
Claude will load the framework, identify what's structurally novel about your problem, and extend the vocabulary as needed. Extension is key: off-the-shelf TopoGlyph rarely fits a specific problem cleanly. The interesting reasoning happens when Claude introduces a few new symbols specifically for your situation.























