lit-scanlisted

# lit-scan Multi-source literature search with a transparent relevance score. Built for tracking an economics literature over time, not just one-off lookups. ## Why not Google Scholar directly Google Scholar has no public API, and its HTML blocks automated access (CAPTCHA after a few programmatic hits, with a real risk of a temporary IP block). So the engine runs on APIs that are stable, free, and give *better* structured metadata than Scholar exposes: OpenAlex (~250M works, topic/field labels), arXiv (econ.* preprints, same-day), Crossref (publisher record of truth), Semantic Scholar, and the NBER working-paper feed. Scholar is still reachable two ways when it is specifically wanted — see [Using Google Scholar anyway](#using-google-scholar-anyway). ## Setup Run once per machine (optional but recommended — it moves OpenAlex and Crossref into their faster, more reliable "polite" pools): ```bash setx LIT_SCAN_MAILTO "you@example.com" ``` No API keys are needed. `SEMANTIC_SCHOLAR_API_KEY` is used if set; without it the `s2` source is rate-limited and is left out of the default source list. ## Quickstart The script lives at `~/.claude/skills/lit-scan/scripts/lit_scan.py`. The standard invocation — top 10, full abstracts to summarise from, score breakdown: ```bash python ~/.claude/skills/lit-scan/scripts/lit_scan.py "AI exposure occupational mobility" --since 2024 --limit 10 --full-abstract --explain ``` Then report the results as specified in [Reporting results](#repo