Generates, validates, and queries Knows research-paper sidecars from papers and scholarly APIs.
npx skills add https://github.com/javiruu/viru-air --skill oma-scholar-javiruu
Please help me install this Agent Skill. Skill: oma-scholar Source: https://github.com/javiruu/viru-air/tree/main/.agents/skills/oma-scholar Command: npx skills add https://github.com/javiruu/viru-air --skill oma-scholar-javiruu
.knows.yaml sidecars that expose a paper's claims, evidence, and relations in a validated, token-efficient format. ## Core Features & Use Cases - Sidecar Generation & Validation: Create v0.9.0 paper@1 sidecars from drafts, LaTeX, or abstracts, then lint them with oma scholar lint in strict or lenient modes. - Remote Search & Fetch: Search knows.academy (~50K indexed papers) with automatic OpenAlex fallback, and fetch full or partial sidecar sections (statements, evidence, relations, artifacts, citation) for up to 93% token savings. - Review, Analyze & Compare: Produce peer-review sidecars, answer natural-language questions over existing sidecars, and structurally diff two papers by claims, methods, and evidence. - Use Case: Ask for the claims of a specific arXiv paper; the Skill resolves it via knows.academy or OpenAlex, fetches only the statements section (~700 tokens), and summarizes each claim with its supporting evidence IDs. ## Quick Start Ask the assistant to search for papers on a topic or generate a .knows.yaml sidecar from your paper draft, for example: "Generate a Knows sidecar from my paper draft and lint it."High-intent search queries and answers about installing and using this skill.
Provide your paper draft, LaTeX, abstract, or research notes and ask for sidecar generation. The Skill extracts claims, evidence, and relations into a v0.9.0 sidecar, then runs oma scholar lint to validate it before reporting counts and any omitted fields.
Run oma scholar search with your query; it queries knows.academy first and automatically falls back to OpenAlex for pre-2026 or non-indexed papers. Use oma scholar resolve for a specific title or oma scholar get with a record ID, DOI, or OpenAlex W-id.
Yes, partial fetch retrieves a single section such as statements, evidence, relations, artifacts, or citation via oma scholar get --section statements. A statements-only fetch is roughly 700 tokens, about 93% smaller than the full PDF.
The v0.9.0 spec requires exact field names like statement_type, evidence_type, predicate, and artifact_type, and actor type must be tool, person, or org, never ai or llm. Numbers must be unquoted and provenance uses a single actor object, not an actors array.
knows.academy indexes only 2026 papers, so older works automatically fall back to OpenAlex for metadata and abstracts. You can then generate a local sidecar from the OpenAlex abstract using Generate mode.
Use --lenient when consuming third-party sidecars fetched from knows.academy, since about 47% contain dangling cross-references. Lenient mode downgrades dangling reference errors to warnings; keep strict mode for your own generated output.
AlterLab-IEUGenerates conference-style posters, flyers and digital signage from a simple text input.
wjgoarxivCoordinate multiple researchers to survey literature across databases and generate survey-report.md.
zhuyingqinVerify LaTeX citations against canonical metadata and cited claim support.
SillyHippyConvert PDF files to Markdown format with preserved structure and extracted images.