llm-wiki

Ingest sources into an interlinked markdown knowledge base with wikilinks and provenance markers.

1|Updated Jun 25, 2026
One-click install
npx skills add https://github.com/Signmanal/VIGIL --skill llm-wiki-signmanal
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: llm-wiki
Source: https://github.com/Signmanal/VIGIL/tree/main/skills/research/llm-wiki
Command: npx skills add https://github.com/Signmanal/VIGIL --skill llm-wiki-signmanal

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill eliminates the inefficiency of traditional RAG systems that re-discover knowledge from scratch for every query, reducing redundant work, missed cross-references, and inconsistent compiled knowledge for research and analysis workflows.

Core Features & Use Cases

  • Persistent Compounding Knowledge Base: Build a self-contained wiki of interlinked markdown files that accumulates knowledge over time with no external database required.
  • Automated Ingestion & Cross-Referencing: Ingest sources including web articles, research papers, and meeting transcripts, then automatically create or update entity, concept, and comparison pages with proper wikilinks and provenance markers.
  • Use Case: For an AI researcher tracking LLM advancements, use this Skill to ingest new arxiv papers, automatically link new model entries to existing architecture and benchmark pages, and maintain a consistent, searchable knowledge base without manual cross-referencing.

Quick Start

Use the llm-wiki skill to ingest the attached arxiv paper on transformer architectures and update all relevant existing wiki pages with new findings.

Frequently Asked Questions about llm-wiki

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

FAQPage Schema
How do I build a persistent markdown knowledge base that compounds research over time?

You can build a persistent markdown knowledge base by maintaining a local wiki directory structured with raw source layers, entity pages, schema conventions, and index files. This interlinked markdown system compounds research over time without requiring an external database.

Why does my research workflow keep re-processing sources and missing cross-references for every query?

Your research workflow re-processes sources because it lacks a persistent knowledge base. Ingesting sources into an interlinked markdown wiki with automated cross-referencing and provenance markers compiles knowledge once, eliminating redundant re-discovery and missed connections across queries.

Can I ingest research papers and meeting transcripts to automatically update entity and concept pages?

Yes, you can ingest sources like research papers and meeting transcripts to automatically create or update entity, concept, and comparison pages. The system applies proper wikilinks and provenance markers to ensure consistent curation and cross-referencing across your knowledge base.

Does this interlinked wiki approach work without an external database?

Yes, this interlinked wiki approach works entirely without an external database. It relies on a self-contained local directory of markdown files structured with raw source layers, schema conventions, and navigation logs to support consistent curation and persistent cross-referencing.

What is the best way to track LLM advancements from arxiv papers without manual cross-referencing?

The best way to track advancements without manual cross-referencing is ingesting arxiv papers into a persistent markdown wiki. This automatically links new model entries to existing architecture and benchmark pages, maintaining a searchable, compounding knowledge base for security analysis and research.

Do I need a specific directory structure to maintain source ingestion and entity tracking?

Yes, you need a local wiki directory structured with raw source layers, entity and concept pages, schema conventions, and index or log navigation files. This specific structure is required to support consistent curation, source ingestion, entity tracking, and cross-referencing.