Skip to content

Codebase Knowledge Graph

Maps a codebase's architecture and relationships into an interactive knowledge graph

Stars
89,744
Licence
MIT
Files
15 files

What it does

When you ask the agent to analyze a project, this skill makes it scan the code, group related files, extract functions, classes, imports and exports with helper scripts, and write a knowledge-graph.json file into a hidden .ua/ (or legacy .understand-anything/) folder. That file powers an interactive dashboard for exploring how the project's components connect. It can also run incremental updates, optionally on each commit, and supports output in other languages.

Great for

  • Developers onboarding onto a large unfamiliar codebase who want an architecture map
  • Teams wanting a browsable overview of components and their relationships
  • Engineers who want incremental graph updates as the code changes

Skip it if

  • Small projects where reading the code directly is quicker (the setup and analysis overhead is not worth it)
  • Users who do not want scripts run on their machine: the skill executes Node.js scripts and may build the plugin on first run
  • Sensitive repositories if the analysis or dashboard is configured to send content externally; the files reviewed do not show the full data path, so check before use

Try asking

  • “/understand”
  • “/understand --full”
  • “/understand ../other-project --language japanese”
  • “/understand --auto-update --exclude "tests/*,docs/*"”
Step-by-step guide: how to use Codebase Knowledge GraphInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 15 original files, the MIT licence and an attribution note.

    understand.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip understand.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip understand.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Run with --full the first time on a large repo, then use incremental updates
  • Use --exclude to skip tests, generated code or docs for a cleaner graph
  • Run from the main checkout rather than a temporary worktree so the graph is not lost when the session ends

What's inside

  • build-fingerprints.mjs
  • compute-batches.mjs
  • extract-structure-result.mjs
  • extract-structure.mjs
  • finalize-incremental.mjs
  • frameworks/django.md
  • frameworks/express.md
  • frameworks/fastapi.md
  • frameworks/flask.md
  • frameworks/gin.md
  • frameworks/nextjs.md
  • frameworks/rails.md
  • frameworks/react.md
  • frameworks/spring.md
  • SKILL.md

Redistributed unmodified from Egonex-AI/Understand-Anything under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.