CloudAI-X

X Bookmarks Skill Graph — Testing skill for Claude Code

Testing community

A skill-graph knowledge system for organizing X (Twitter) bookmarks as an interconnected graph of ideas.

How to install X Bookmarks Skill Graph

This entry records only its repository, not the path inside it, so there is no exact command to give. Open CloudAI-X/x-bookmarks-skill-graph and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What X Bookmarks Skill Graph does

A skill-graph knowledge system for organizing X (Twitter) bookmarks as an interconnected graph of ideas. Testing graph-native knowledge architecture vs traditional SKILL.md files.

Alternatives in Testing

Claude Code Spec Workflow — Automated Kiro-style Spec workflow for Claude Code 3.6k ★
Langgraphjs — by langchain-ai - Offers comprehensive build and test commands with detailed TypeScript style guidelines, laye 2.7k ★
Spec Driven Develop — Spec-driven development workflow for AI coding agents: architecture-first planning, task decomposition, GitHub 975 ★

README

X Bookmarks Knowledge Graph

A **skill-graph knowledge system** for organizing X (Twitter) bookmarks as an interconnected graph of ideas, not a flat folder of files.

This repository is an experimental testbed comparing **skill-graph architecture** (interconnected markdown files with wikilinks) against traditional **SKILL.md** files for agent knowledge management.

Purpose

This project exists to answer one question:

**How does a skill-graph knowledge system compare to SKILL.md files for agent knowledge retrieval and reasoning?**

Traditional agent skills use monolithic `.md` files with structured frontmatter. This repository tests an alternative: a **graph-native** approach where:

Every bookmark is a node in a knowledge graph
Connections are explicit via related: links in YAML frontmatter
Navigation happens through bidirectional links, not hierarchical folders
Concept MOCs (Maps of Content) serve as entry points for traversal
Agents traverse the graph by following pointers, not scanning directories

Architecture

Core Principles

Principle	Implementation
Graph-native	Every bookmark is a node; `related:` links form edges
Bidirectional links	If A links to B, B must link back to A
Progressive disclosure	Read frontmatter (15 lines) → follow `related:` → read full content only when needed
Concept MOCs	Curated entry points for topic-based exploration
YAML frontmatter	Machine-readable metadata for agent traversal