An entire scoping review in one tool — and a living one: every step re-runs on demand, updating results without redoing human work. Query building with LLM source-aware translation to 14 databases (direct harvest for the four with free APIs), cross-database deduplication, blind multi-reviewer screening with conflict resolution, full-text retrieval ladder, structured extraction with AI first-pass drafts, procedural citations, live PRISMA diagram, and a public read-only dashboard per review. A read-only MCP surface lets a model client read a review from inside a conversation — records, votes, criteria, extraction, synthesis — with a per-user key that reaches only what its owner reaches.
Who it is for: A review team working the same corpus — screeners, extractors, and whoever reads the PRISMA at the end.
Source code: https://github.com/that-ugly-cat/lssr