aris-research-wiki

Maintains a persistent research knowledge base of papers, ideas, experiments, and claims.

1.1k|116|Updated Feb 26, 2026
One-click install
npx skills add https://github.com/OpenLAIR/dr-claw --skill aris-research-wiki
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: aris-research-wiki
Source: https://github.com/OpenLAIR/dr-claw/tree/main/skills/aris-research-wiki
Command: npx skills add https://github.com/OpenLAIR/dr-claw --skill aris-research-wiki

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Research projects lose accumulated knowledge across sessions: literature surveys are forgotten, failed ideas get retried, and experiment results are disconnected from the claims they tested. This Skill builds a persistent, per-project research wiki that compounds knowledge across the entire research lifecycle.

Core Features & Use Cases

  • Structured Knowledge Graph: Stores papers, ideas, experiments, and claims as typed entities with relationships (extends, contradicts, supports, invalidates) materialized in a graph/edges.jsonl file.
  • Compressed Query Pack: Generates a hard-budgeted 8000-character query_pack.md summarizing top gaps, paper clusters, and failed ideas for downstream ideation workflows.
  • Health Checks and Stats: The lint subcommand detects orphan pages, stale claims, contradictions, and dead ideas, while stats reports entity counts and gap status.
  • Use Case: After running a literature survey, ingest the top papers into the wiki, then let the idea-generation workflow read the query pack so it avoids previously failed ideas and targets unresolved field gaps.

Quick Start

Initialize a research wiki for this project and ingest the paper "Attention Is All You Need" from arXiv 1706.03762.

Frequently Asked Questions about aris-research-wiki

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

FAQPage Schema
How do I build a persistent research knowledge base for my project?

Run the init subcommand to create a research-wiki directory with papers, ideas, experiments, claims, and a graph folder. Then ingest papers with metadata fetched from arXiv, DBLP, or Semantic Scholar, and the wiki maintains indexes, gaps, and a query pack automatically.

How do I track relationships between research papers and experiments?

Relationships are stored as typed edges in graph/edges.jsonl, including extends, contradicts, inspired_by, tested_by, supports, and invalidates. Each entity page has an auto-generated Connections section derived from this graph, so edges.jsonl remains the single source of truth.

What is the query pack in a research wiki used for?

The query pack is a compressed summary capped at 8000 characters covering project direction, top gaps, paper clusters, failed ideas, and open questions. Ideation workflows read it before generating new ideas so failed approaches are treated as a banlist and unresolved gaps as search seeds.

Can the research wiki detect contradictions or stale claims?

Yes, the lint subcommand health-checks the wiki for orphan pages, claims with conflicting supports and invalidates edges, stale reported claims older than 14 days, dead untested ideas, and sparse pages. It outputs a LINT_REPORT.md with suggested fixes.

When should I re-run idea generation after updating the wiki?

Re-ideation is suggested after five or more new papers are ingested, three or more ideas fail or partially fail, a new contradiction appears, or a new gap has no addressing idea. The system suggests re-running but never triggers it automatically.