kmshihab7878 /understand-setup
Analyze the Claude Code setup at ~/.claude/ and produce an interactive knowledge graph showing how skills, agents, commands, MCP servers, docs, and memory all connect.
Instructions
cd ~/.claude/tools && python3 -m coremind_understand ~/.claude --mode claude
Then launch the dashboard:
cd ~/Projects/understand-anything/understand-anything-plugin/packages/dashboard && GRAPH_DIR=~/.claude npx vite --open
What it maps
- Skills (48): Each SKILL.md becomes a file node with category tags
- Agents (39+): Each agent definition becomes a file node
- Commands (64+): Each command definition becomes a file node
- Docs (16): Reference documents become file nodes
- Memory (5): Persistent memory files become file nodes
- MCP Servers (8+): Each server becomes a concept node
- Tools: Custom tool packages become file nodes
Layers
The graph is organized into 7 layers:
- Skills — Skill definitions
- Agents — Agent definitions
- Commands — Command definitions
- Documentation — Reference docs
- Memory — Persistent memory
- Tools — Custom packages
- Configuration — Config files
Tour
8-step guided tour: Overview → Skills → Agents → Commands → MCP Servers → Memory → Operating Framework → Operating Framework