Pick one of the six projects in this course and write its full spec before any code: claim.md with the blanks unfilled, done_when.md with a five-minute verification path a stranger could follow, a corpus smell-test result, and a cost estimate from a 10-document sample. Submit the four files plus one paragraph on why you chose that project over the others.
$ python corpus_smell_test.py ./data/filings
{'docs': 420,
'pct_docs_sharing_an_entity': 78.3,
'vocab_depth': 0.214,
'verdict': 'usable'}
$ python estimate.py --sample 10
per doc ~ $0.0412 | 420 docs ~ $17.30
ceiling set: MAX_SPEND_USD = 25.00