Windows-first、local-first AI 科研文献工作台：支持 Zotero 增量同步、DOI/开放 PDF 获取、MinerU/PyMuPDF 解析、MiniMax 向量化、混合检索、DeepSeek 证据问答、在线文献发现、引用导出与 MCP 集成，无需 WSL、Docker 或 PostgreSQL。
Skill Claude CodeCodex
Normalize and batch-import DOIs into ResearchBrain, resolve bibliographic metadata, find lawful open-access PDFs, store them locally by content hash, and track parsing and embedding. Use when Codex receives one or more DOIs, needs to obtain available PDFs, must avoid duplicate literature, or must report whether each…
Skill Claude CodeCodex
Conduct evidence-grounded literature research by having Codex directly search and synthesize ResearchBrain full-text evidence, then extend coverage with Crossref, OpenAlex, arXiv, and PubMed metadata. Use for literature reviews, method and dataset comparisons, research-gap analysis, DOI-backed reports, paper-specific…
Skill Claude CodeCodex
Use ResearchBrain local libraries and academic discovery tools for evidence-grounded literature research.
Skill Claude CodeCodex
Attach user-selected local PDFs or process ResearchBrain-managed PDFs into deduplicated objects, structured Markdown, document artifacts, chunks, and vectors using MinerU with PyMuPDF fallback. Use when Codex is asked to import a PDF, convert PDF to Markdown, parse Zotero attachments, resume failed document…
Skill Claude CodeCodex
Build, complete, and audit ResearchBrain vector indexes for bibliographic metadata, abstracts, and parsed PDF full text. Use when Codex is asked to vectorize a library, repair missing embeddings, check index coverage, process only incremental gaps, or verify that literature is searchable.
Skill Claude CodeCodex
Incrementally synchronize a local Zotero library into a ResearchBrain mirror, including metadata changes, deletions, linked PDF attachments, parsing, and downstream indexing. Use when Codex is asked to import newly added Zotero papers, sync only changes since the last run, diagnose missing Zotero items or PDFs, or…