Knowledge Graph Analyzer (understand-knowledge)

Scans a Karpathy-pattern markdown wiki and builds an interactive knowledge graph with entities, implicit relationships, and topic clusters.

Data & AnalyticsIntermediate★ 78,314⑂ 6,578AI score 7/10Last updated: Jul 30, 2026

What it does

It analyzes a markdown wiki that uses index.md plus [[wikilinks]] — Karpathy's three-layer LLM wiki pattern (raw sources + generated wiki + schema file) — and produces a knowledge graph JSON.

The workflow is a five-phase pipeline:

DETECT — runs the bundled parse-knowledge-base.py to confirm the wiki structure and report article, source, topic, and wikilink counts.
SCAN — deterministically extracts wikilinks, headings, frontmatter, and index.md categories into scan-manifest.json.
ANALYZE — dispatches article-analyzer subagents in batches of 10–15 articles (up to 3 concurrently) to infer implicit relationships, entities, and claims that explicit links miss.
MERGE — deduplicates entities, normalizes node/edge types, and builds layers plus a guided "tour" from index.md ordering.
SAVE — validates and drops dangling edges, writes knowledge-graph.json and meta.json, then auto-invokes the dashboard skill.

Notably, article content is explicitly treated as untrusted data (prompt-injection aware) and the cleanup step guards the rm -rf path against empty variable expansion.

Who it's for

Obsidian / Zettelkasten / Karpathy-wiki users with hundreds of wikilinked notes
Researchers who want to surface hidden connections across saved papers and articles
Documentation owners auditing a team knowledge base for orphan notes, broken links, and topic gaps

Examples

/understand-knowledge ~/notes/ai-wiki — scans 250 notes and reports "250 articles, 40 sources, 18 topics, 1,120 wikilinks (34 unresolved)" before building the graph.
Surfaces an implicit edge between your reinforcement-learning notes and robotics notes that were never directly linked, exposing a gap in your research map.
Uses the generated tour steps (ordered by index.md sections) as an onboarding reading path for a new teammate.

· · · Install guide · · ·

Install in the Claude app (no terminal)

Download the ZIP with the button below.
In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.

↓ Download ZIP

Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the understand-anything-plugin/skills/understand-knowledge folder from the GitHub repo Egonex-AI/Understand-Anything into my ~/.claude/skills/understand-knowledge/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/Egonex-AI/Understand-Anything.git /tmp/understand-anything && cp -r /tmp/understand-anything/understand-anything-plugin/skills/understand-knowledge ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

Open a terminal and confirm prerequisites with python3 --version and git --version.
Clone the repository into a temp folder: git clone https://github.com/Egonex-AI/Understand-Anything.git /tmp/understand-anything
Copy the skill into your Claude Code skills folder: mkdir -p ~/.claude/skills && cp -r /tmp/understand-anything/understand-anything-plugin/skills/understand-knowledge ~/.claude/skills/
Also copy the sibling understand-dashboard skill from the same plugin, since this skill auto-triggers it to visualize the graph.
Verify that ~/.claude/skills/understand-knowledge/ contains SKILL.md, parse-knowledge-base.py, and merge-knowledge-graph.py.
Restart Claude Code and run /understand-knowledge ~/path/to/your/wiki.
Output lands in .ua/knowledge-graph.json inside your wiki folder (or .understand-anything/ if that legacy directory already exists).

View source on GitHub ↗License: MIT