# Graph Thinking

> Model complex systems as nodes, edges, flows, and dependencies. Use when mapping stakeholder networks, causal loops, product ecosystems, architecture dependencies, knowledge graphs, or non-linear strategy problems; avoid when a simple hierarchy, roadmap, or checklist is enough.

- Skill: `flpbalada/graph-thinking` (Agent Skill)
- Install (CLI): `npx skillmds@latest add flpbalada/graph-thinking`
- Raw SKILL.md: https://api.skillmd.com/api/skills/flpbalada/graph-thinking/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Product & Planning, Coding & Dev Tools
- Author: flpbalada (https://skillmd.com/u/flpbalada)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/flpbalada/graph-thinking

---


# Graph Thinking

Model a problem as nodes and relationships.
Use it when linear lists hide dependencies.

## Goal

- Identify important nodes.
- Name relationships.
- Find clusters, bottlenecks, loops, and paths.
- Turn structure into decisions.

## Concepts

- Node: item, person, system, idea, or constraint.
- Edge: relationship between nodes.
- Cluster: tightly connected group.
- Path: route through graph.
- Centrality: importance by connectedness.
- Loop: circular dependency or feedback.

## Edge types

- Dependency: A requires B.
- Influence: A affects B.
- Conflict: A competes with B.
- Synergy: A improves B.
- Flow: A sends value, data, or work to B.

## Rules

- Do not connect everything to everything.
- Name edge types.
- Mark edge strength when useful.
- Look for missing nodes.
- Look for critical paths.
- Prefer simple diagrams over dense tables.

## Flow

1. Define subject and decision to support.
2. List nodes.
3. Group nodes by category.
4. Add typed edges.
5. Identify central nodes, clusters, loops, and bottlenecks.
6. Trace key paths.
7. Recommend actions.

## Resources

- [Graph of Thoughts: ArXiv Paper](https://arxiv.org/abs/2308.09687)
- [Neo4j Graph Database Use Cases](https://neo4j.com/use-cases/)
- [Network Science by Albert-László Barabási](http://networksciencebook.com/)

## Output

```md
## Graph Analysis

Subject: [what is being mapped]

## Nodes
- [category]: [nodes]

## Relationships
- [A] -> [B]: [edge type], [strength]

## Patterns
- Central nodes: [nodes]
- Clusters: [groups]
- Bottlenecks: [nodes]
- Loops: [loops or none]
- Critical paths: [paths]

## Recommendations
- [action]
- [action]
```