← Back to Discover

Understand Anything

Open-source plugin that analyses a codebase or LLM wiki with a multi-agent pipeline and builds an interactive knowledge graph you can explore, search and question, inside Claude Code, Codex, Cursor and others.

Coding agentsNotes & knowledge
Visit website ↗

What it does

Understand Anything analyses a project and builds a knowledge graph of its files, functions, classes and dependencies, then opens an interactive dashboard to explore it. It is delivered as a plugin for Claude Code and can also be installed for Cursor, VS Code with GitHub Copilot, Copilot CLI, Codex, Gemini CLI, OpenCode and many other assistants. Static parsing with tree-sitter handles structure such as imports and definitions, and language models add summaries, architectural layers, guided tours and business-domain mapping. The main command runs a pipeline of specialised agents, and later runs only re-analyse files that changed. The README warns that the first analysis of a large project can use a significant number of tokens and suggests a subscription plan or a local model for that run. Other commands answer questions about the codebase, explain a single file, show the impact of current changes and generate an onboarding guide. A separate command turns a Karpathy-pattern LLM wiki into a knowledge graph. Teams can commit the generated graph and open the dashboard with only Node.js, without an LLM or an API key.

Capabilities

  • Interactive knowledge graph of files, functions, classes and dependencies
  • Guided tours of the architecture, ordered by dependency
  • Fuzzy and semantic search across the graph
  • Diff impact analysis before you commit
  • Domain view that maps code to business flows
  • Knowledge-base analysis for LLM wikis

When to use it

  • You have joined a team with a large, unfamiliar codebase
  • You want to see which parts of the system a change affects
  • You want an onboarding guide for new team members
  • Teammates should browse a shared graph without running an LLM