Subagent-Driven Literature Review

RC native tools — this skill's pipeline ships as four built-in RC tools. Call them directly; do not shell out to any rp.py script or write curl:

• rp_search({ query, limit?, min_year? }) — Scopus relevance search enriched with OpenAlex abstracts + OA PDF links. The Elsevier key is built in.
• rp_abstracts({ dois }) — batch abstracts + OA links for a DOI list (OpenAlex, no key).
• rp_cite({ doi, direction?, limit? }) — citation traversal (direction: both / backward / forward).
• rp_fulltext({ doi, out? }) — OA full text (Elsevier ScienceDirect OA → OpenAlex OA fallback); pass out to also save the text to a file.

Results return inline as JSON (there is no --json <file> flag). To persist a result set, save the returned JSON with the workspace file tools.

Overview

Core principle: Fresh subagent per batch + consolidation between batches = parallel reasoning with quality control.

For large reviews (50+ papers), subagents help in two real ways: