Run the daily arXiv workflow for this project: fetch a date/category batch, recommend papers against the local user profile, perform selective close reading, check missing citations, use cautious gated research-idea generation, and write user-facing reports…
Use when the user asks substantive follow-up questions about a specific paper or asks to discuss, explain, critique, or connect a paper to their research. Maintain one evolving local note per paper in data/notes/ARXIV_ID.md, read local notes and source before…
Use proactively whenever the user gives feedback on a paper or idea, or engages in sustained discussion after a recommendation: record sparse feedback, update idea status, and promote stable patterns into the durable research profile.
Initialize or refresh this project's private user profile from Google Scholar or user-provided publication data, filter non-article records, and produce a research-interests profile through evidence-based synthesis and targeted user questions.