What is the Reflect skill in Claude Code?

The Reflect skill is a specialized tool that analyzes text files to identify conceptual relationships, creating a connected knowledge graph by updating Maps of Content (MOCs) and adding bidirectional links.

What is the 'Articulation Test' used by this skill?

The Articulation Test is a quality control mechanism where the skill must be able to explicitly state *why* two notes relate (e.g., 'X extends Y' or 'A contradicts B') before a connection is approved.

Does this skill require specific directory structures?

While it has defaults, it is highly configurable via a derivation-manifest.md file, allowing you to define custom folder names for notes, inboxes, and topic maps.

How does the Reflect skill find connections?

It uses a dual-discovery method that explores existing topic maps while simultaneously running semantic vector searches to find conceptual neighbors that might not share the same keywords.

Can Reflect perform semantic search without external tools?

It prefers MCP tools like mcp__qmd for deep search but features a robust fallback system that can utilize bash-based queries or standard grep if specialized indexing tools are unavailable.

Knowledge Graph Reflector

Name: Knowledge Graph Reflector
Author: LopeWale

byLopeWale

0•

生産性とワークフロー

Identifies semantic relationships between notes and automatically updates Maps of Content (MOCs) to build a connected knowledge graph.

The Reflect skill transforms isolated documentation and notes into a cohesive knowledge system by identifying deep conceptual connections and synthesizing them into structured Maps of Content. Using a dual-discovery approach that combines existing structural navigation with multi-tier semantic search, it ensures every link is meaningful and articulable. It is an essential tool for researchers, developers, and writers using Claude Code to maintain high-quality project wikis, ensuring that new insights are integrated into the existing body of work rather than remaining in silos.

主な機能

01Semantic judgment engine that requires an 'Articulation Test' for every connection

02Dual-discovery linking combining MOC exploration and semantic vector search

03Multi-tier search fallback supporting MCP tools, bash-based qmd, and grep

04Automated Map of Content (MOC) synthesis and cross-linking

050 GitHub stars

06Discovery Trace logging for transparent and reproducible methodology

ユースケース

01Synthesizing disparate research notes into a centralized topic map

02Finding hidden connections between technical documents and architecture decisions

03Maintaining a 'Zettelkasten' style digital garden within a software repository

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lopewale/ampllabs reflect

For use in Claude.ai and ChatGPT