skills.rest
tronghieu
tronghieuCOMMUNITY
Explore all skills by this author →
GitHub

lumi-research-discover

Discover and rank candidate papers or web sources for a research topic.

36|16|Updated May 1, 2026
Search Tags:#research#citations#deduplication#ranking#shortlist#source discovery#metadata-first
One-click install
npx skills add https://github.com/tronghieu/lumina-wiki --skill lumi-research-discover
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: lumi-research-discover
Source: https://github.com/tronghieu/lumina-wiki/tree/main/src/skills/packs/research/discover
Command: npx skills add https://github.com/tronghieu/lumina-wiki --skill lumi-research-discover

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Helps you quickly find high-quality candidate sources for a research topic and reduce time spent searching before you commit to ingestion.

Core Features & Use Cases

  • Source discovery with shortlist approval: Proposes and ranks candidate papers or web sources, then stops for user review instead of ingesting automatically.
  • Deduplication and checkpointing: Avoids duplicates by comparing against existing wiki sources, discovered JSON, and discovery checkpoints.
  • Purpose-aware ranking: Evaluates how well each candidate aligns with the repository’s stated project purpose and flags off-purpose items.

Quick Start

Ask: “Use /lumi-research-discover to discover and rank a shortlist of sources for my topic: <topic>, dedupe against what I already ingested, and ask me which to ingest.”

Frequently Asked Questions about lumi-research-discover

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

FAQPage Schema
How do I find and rank research sources before committing to full PDF ingestion?▼

Source discovery tools find candidate papers or web sources for a research topic, rank them by purpose alignment, and produce a reviewable metadata shortlist for approval before any PDF ingestion or wiki updates occur.

What's the best way to avoid duplicate research papers when discovering new sources?▼

Deduplication during source discovery compares new candidate papers against existing wiki sources, discovered JSON files, and discovery checkpoints to prevent duplicate research sources from entering your repository.

Can I review candidate research sources before they are automatically added to my wiki?▼

Yes, metadata-first source discovery proposes and ranks candidate papers, then stops for user review and shortlist approval instead of automatically ingesting sources or mutating existing wiki state.

How does purpose-aware ranking evaluate candidate papers for a research repository?▼

Purpose-aware ranking evaluates how well each candidate source aligns with your repository’s stated project purpose, ranking relevant papers higher and flagging off-purpose items for exclusion from the shortlist.

Do I need to provide specific input formats to start discovering research sources?▼

You need to provide a research topic and have reading discovery setup and source-mode references available, allowing the tool to deduplicate and write candidate metadata JSON under raw/discovered without mutating wiki.

What limitations exist with metadata-only source discovery for research workflows?▼

Metadata-only source discovery strictly limits the workflow to finding and ranking candidate papers, writing candidate metadata JSON without ingesting PDFs, updating wikis, or mutating any existing repository state.

Related Skills

Explore More Education & Research Skills→
AvinavaAvinava
1

sf-knowledge-clm-developer-guide

Provides curated Salesforce Contracts knowledge for configuring, customizing, and extending contract capabilities via indexed topic files and API references.

Community
Intermediate
xiaonanjixiaonanji

person-wiki-knowledge

Search and ingest knowledge from the person-wiki repository using wiki.py.

Community
Intermediate
tronghieutronghieu
36

lumi-research-watch-run

Run scheduled watchlist discovery and return a JSON summary of new paper candidates.

Community
Intermediate
smallnestsmallnest
190

pdf

Extract text and tables from PDFs with layout awareness.

Community
Advanced