Foundation Due Diligence

Problem

Foundation due diligence is hard not because information is scarce, but because public evidence is fragmented: PDF annual reports provide authoritative but lagging statutory figures, website slideset images carry the latest operational data, and project subpages contain spend, beneficiary, outcome, and grant details. Scraping only the public-information homepage can miss the latest fiscal year, create inconsistent amounts, leave project evidence vague, and skip subpages, making it difficult to use the report for follow-up verification.

How It Works and Limits

The skill takes a foundation or social organization name and runs four steps: Step 0 gathers public data, detects WeZhan-style sites, crawls columns and sitemap.xml, extracts Body.js, downloads and OCRs images, and consolidates scattered financial records by fiscal year; Step 1 maps the data into 69 checkpoints, separates statutory and dynamic data sources, records the data-mapping, and limits unresolved items; Step 2 generates a structured Markdown report; Step 3 checks fiscal-year continuity, units, sources, and amount consistency. It fits engineering workflows that need a diligence draft, a compliance snapshot, and a public-information checklist, as well as cases requiring traceability, review, and batch comparison of public data. It does not replace legal advice, audit conclusions, or regulatory verification, and results depend on site accessibility, image legibility, and disclosure completeness.