# Refresh the Research Desk

Work in `/Users/mr.adams/financial-research-monitor`. Read README and CURATION.md first. The user wants selective, current, best-in-class information across the 91 existing providers. Do not restart the retired bulk archive crawl or pursue a link-count target.

Review recent official publications and the small review queue. Prefer the past 30 days, then 90 days, and the latest useful annual research. Read source material before selecting. A famous publisher or an interesting title is not enough.

Select only work with original evidence, transparent methods, concrete implementation lessons or a material market/policy change. Explain why it matters, the client decision it informs, the data period, and limitations. Label public-summary-only reviews and gated reports. Verify publication dates; never substitute crawl dates or sitemap modification dates. Compare commercial perspectives with independent or official evidence. Distinguish announced, piloted and operational capabilities, and proposed versus final rules.

Recommend up to five current reads for the daily brief, with sector diversity when justified. Do not fill a quota with weak material. Update `must_read.json` only with evidence-backed selections and actual review dates. Retain older items only for a stated continuing use. Record replaced annual editions explicitly; do not silently retitle an existing URL.

Run `python3 curation.py --check`, the relevant tests and local browser checks. Export through `python3 monitor.py export`. Publication authority applies only to the existing `sources-mradams` research dashboard; preserve other projects, DNS and local source data. Use the guarded `monitor.py publish` helper and verify the live site. No purchases, subscriptions, credentials, outreach or paywall bypass.

Report what earned selection, what was excluded and why, which priority publishers were inaccessible, and the dates actually covered. Automatic monitoring and keyword screening do not constitute editorial review.