Verified:GitHub

lumi-ask

Retrieves cited answers from Lumina-Wiki's graph of concepts and sources.

36|16|Updated May 1, 2026

Search Tags:#knowledge retrieval #gap analysis #research assistant #graph traversal #question answering #cited synthesis #wiki indexing

Ready to run

npx skills add https://github.com/tronghieu/lumina-wiki --skill lumi-ask

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: lumi-ask
Source: https://github.com/tronghieu/lumina-wiki/tree/main/src/skills/core/ask
Command: npx skills add https://github.com/tronghieu/lumina-wiki --skill lumi-ask

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you answer questions using only what your Lumina-Wiki already knows, avoiding guessing, unsupported claims, or re-reading raw sources.

Core Features & Use Cases

Wiki-graph retrieval: Traverses the wiki index, edges, and citations to gather a relevant subgraph of related concepts and sources.
Cited synthesis: Produces a direct answer first, then backs it with verifiable citations to wiki pages.
Gap detection: Clearly reports what the wiki does not cover and suggests using /lumi-ingest to fill missing knowledge.
Optional filing workflow: Can save a summarized result to wiki/outputs or wiki/summary only after explicit user confirmation.

Quick Start

Ask a question like: "What does the wiki say about flash-attention variants, and can you compare them across the most recent sources?"

Frequently Asked Questions about lumi-ask

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

FAQPage Schema

How do I get cited answers from a wiki knowledge graph instead of guessing?▼

To retrieve cited answers from a wiki knowledge graph, you query already-ingested content to traverse concepts and sources. The system synthesizes a direct answer backed by verifiable citations to wiki pages, preventing unsupported claims.

How does question answering work when traversing a wiki index and edges?▼

Question answering across a wiki index and edges works by extracting a relevant subgraph of concepts and sources using read-only subcommands. It returns a synthesized answer grounded in verifiable citations from the wiki pages.

Can I compare concepts across recent sources using wiki graph retrieval?▼

Yes, you can compare concepts across recent sources using wiki graph retrieval. By traversing the graph of concepts, edges, and citations, the system gathers related sources and synthesizes a direct comparison grounded in those specific wiki pages.

What happens when my wiki knowledge graph has coverage gaps during question answering?▼

When your wiki knowledge graph has coverage gaps during question answering, the system clearly reports what is missing. It detects these gaps and suggests using an ingestion command to fill the missing knowledge.

Do I need raw sources to perform question answering and finding related concepts?▼

No, you do not need raw sources to perform question answering and finding related concepts. The system traverses the wiki index, edges, and citations to gather a relevant subgraph, avoiding the need to re-read raw sources.