zhihaoairobotic/clawphd/arxivterminal

arXivTerminal

CLI tool for managing arXiv papers with local database storage.

Dependency note: arxivterminal pins arxiv<2.0.0 and pydantic<2.0.0, but both must be overridden.

Install with: pip install arxivterminal && pip install 'arxiv>=2.0' 'pydantic>=2.0'

Quick Reference

Fetch Papers from arXiv

arxiv fetch --num-days N --categories CATEGORIES

This is non-interactive and safe to call directly. See arxivterminal-fetch.md for details.

Search Local Database

Use the bundled non-interactive search script (the arxiv search CLI is interactive and will hang):

python SKILL_DIR/scripts/arxiv_search.py "QUERY" -l 10

Replace SKILL_DIR with the directory containing this SKILL.md (derive from the <location> field).

Database Statistics

arxiv stats

Database Management

• arxiv delete-all — clear all papers
• See arxivterminal-management.md for database location and backup

Data Storage

Paths are platform-dependent (uses appdirs):

• Linux: ~/.local/share/arxivterminal/papers.db
• macOS: ~/Library/Application Support/arxivterminal/papers.db

Built-in agent tools (no local DB)

For programmatic date-range + topic fetch, scoring, and digests inside ClawPhD, use the three tools documented in the `paper-scout` skill:

• arxiv_fetch_range — Atom API, optional keyword OR-query + categories + submittedDate window
• arxiv_rank_papers — enhanced metadata heuristic + Semantic Scholar/OpenAlex enrichment + optional batch LLM refinement
• arxiv_paper_digest — introduction report for the selected papers using ranking rationale when available

Use arxivterminal when you need a persistent SQLite corpus (arxiv fetch, arxiv_search.py). Use paper-scout tools when you want a one-shot pipeline without populating the DB.

Common Workflows

Daily Research Workflow

arxiv fetch --num-days 1 --categories cs.AI,cs.CL
python SKILL_DIR/scripts/arxiv_search.py "large language models" -l 20

Weekly Review

arxiv fetch --num-days 7 --categories cs.AI,cs.LG,cs.CV
arxiv stats