What problem does it solve? Keyword-based paper search APIs like arXiv and Semantic Scholar often miss relevant work that uses different terminology or spans sub-problems. This Skill uses Gemini to decompose a research topic into multiple angles, aliases, and neighboring tasks, surfacing papers that traditional index searches overlook. ## Core Features & Use Cases - Multi-angle literature discovery: Gemini decomposes the query into sub-problems and naming variants to find genuinely relevant papers across venues, surveys, and preprints. - Structured paper output: Returns title, authors, year, venue, arXiv ID, DOI, code URL, and a one-sentence summary for up to 15 papers, formatted as a table. - MCP-first with CLI fallback: Prefers the gemini-mcp-tool integration and falls back to the gemini CLI with a 120-second timeout when MCP is unavailable. - Use Case: A researcher exploring a new topic like "speculative decoding for LLM inference" runs the search to get a broad paper list with code links, then cross-verifies key results via Semantic Scholar. ## Quick Start Ask the assistant to run a gemini search on your research topic, for example by saying: use gemini search to find recent papers on retrieval-augmented generation evaluation.
