scholar-megasearchlisted

Massive multi-source academic literature search via subagent orchestration. Fans out parallel searchers across every available scholarly source — arXiv, Semantic Scholar, Crossref, OpenAlex, PubMed/PMC, bioRxiv/medRxiv, DOAJ, CORE, BASE, DBLP, IACR, SSRN, Zenodo, Unpaywall, plus web/GitHub — then deduplicates by DOI/arXiv-id/title into one ranked corpus and synthesizes it. Use when the user wants a broad/exhaustive literature sweep, a large-scale paper search, a systematic review corpus, citation snowballing, or to find as many papers as possible on a topic across many databases at once. Triggers: "massive literature search", "literature review", "search across every database", "systematic search", "mega search", "search every source", "exhaustive search". Localized trigger phrases in other languages map to the same intent.
TaewoooPark/scholar-megasearch · ★ 7 · AI & Automation · score 78

Install: claude install-skill TaewoooPark/scholar-megasearch

# scholar-megasearch Integrates every academic search MCP/skill in this environment into one fan-out → merge → synthesize pipeline. Each subagent owns one **source bucket** and searches in parallel; results are merged into a single deduplicated, provenance-tracked, ranked corpus. Prefer this over single-source searches whenever breadth matters. For the full source list and which tools live in each bucket, read `references/sources.md`. For the orchestration templates and the record schema, read `references/orchestration.md`. ## Workflow ### 1. Frame the query + pick a depth level Restate the topic in one line. If it is underspecified (e.g. "find papers on neural networks"), ask 1–2 clarifying questions (sub-aspect? time range? methods vs. phenomena?) before fanning out — a vague seed wastes a large fan-out. Then pick a **depth level L1–L5** (see `## Depth levels` below): it sets the facet count, bucket count, per-source hit cap, and how many waves run. An explicit `depth=N` / `LN` / bare `1–5` in the request wins; otherwise infer from phrasing; otherwise default **L2**. ### 2. Decompose into facets + route to buckets - **Facets** (count set by the depth level, 3–8): synonyms, sub-aspects, method vs. phenomenon, key authors, and at least one each of a broad and a narrow phrasing. For topics with strong non-English literature, add a localized query in the relevant language for Bucket G. - **Buckets** (count set by the depth level, 4–7): pick from the domain→bucket rou