We build open-source tools that make academic research consumable by AI workflows — fetching, compressing, and indexing papers so language models can actually use them.

arXiv papers as clean Markdown for LLMs. 97% smaller than PDF.

Works as a CLI, MCP server, or Claude Code skill. Supports arXiv, PubMed, DOI, and Semantic Scholar out of the box.

npx paper7 get 2312.00752
npx paper7 search "retrieval augmented generation" --limit 5
npx paper7 kb ingest 2312.00752 1706.03762

TypeScript SDK for arXiv/PubMed research, built with Effect v4. Typed services, no implicit side effects.

• LLM-first — output optimized for context windows, not human reading
• Zero bloat — no heavy dependencies, no auth walls, no subscriptions
• Composable — works standalone, in pipelines, or as MCP tools