One-click install
npx skills add https://github.com/Flux-Frontiers/doc_kg --skill kgrag
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: kgrag
Source: https://github.com/Flux-Frontiers/doc_kg/tree/main/.claude/skills/kgrag
Command: npx skills add https://github.com/Flux-Frontiers/doc_kg --skill kgrag

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

KGRAG removes the friction of managing multiple knowledge graphs by unifying registry, search, and snippet extraction into a single workflow you can trust and repeat.

Core Features & Use Cases

  • Unified registry: Track CodeKG, DocKG, and MetaKG instances in one place (by default ~/.kgrag/registry.sqlite), including paths and build status.
  • Federated querying: Run one query across all registered KGs and get globally ranked results.
  • LLM-ready snippet packs: Extract source-grounded code/doc excerpts into a single markdown context file for analysis or review.
  • Visualization & diagnostics: Use an interactive Streamlit UI to inspect registry contents and run cross-KG analysis.

Quick Start

Initialize your repository for unified graph search by running: kgrag init ~/repos/myproject.

Frequently Asked Questions about kgrag

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I query multiple knowledge graphs at once for cross-project code discovery?

You can perform federated querying across multiple knowledge graphs by registering CodeKG, DocKG, and MetaKG instances in a unified registry, then running a single query to retrieve globally ranked results from all registered graphs.

What is the best way to extract source-grounded code snippets for LLM context?

Extracting source-grounded snippet packs involves querying your registered knowledge graphs to pull code and documentation excerpts into a single markdown file, providing grounded context for LLM analysis or architectural review.

Does kgrag integrate with Claude Code via MCP for architectural analysis?

Yes, kgrag integrates with Claude Code and Claude Desktop via MCP using configurable environment variables, allowing you to perform architecture analysis and knowledge graph discovery directly within your AI assistant workflow.

How do I initialize a repository for unified knowledge graph search?

To initialize a repository for unified graph search, run the command `kgrag init ~/repos/myproject`, which sets up the project for tracking within the default registry database located at `~/.kgrag/registry.sqlite`.

Can I filter federated knowledge graph search results by specific component types?

Yes, federated search supports kind-based filtering, allowing you to narrow down query results across registered CodeKG, DocKG, and MetaKG instances to target specific component categories during architectural analysis.

What are the limitations of using a local SQLite registry for knowledge graph management?

The knowledge graph registry uses a local SQLite database by default, meaning registry tracking, path management, and build status checks are confined to a single local environment without built-in remote synchronization capabilities.