22d • Question of the Day
06/23 Question of the Day
A Data Extraction agent instructions include: 'Do not extract data from the header or footer sections of documents.' Extraction results still include header data. Which approach most reliably enforces the exclusion?
A. Repeat the exclusion instruction three times in the system prompt
B. Preprocess documents to strip headers and footers before passing to Claude
C. Wrap header and footer content in <excluded_section> XML tags before injecting the document
D. Lower temperature to make extraction more conservative and precise
Drop your answer (A / B / C / D) in the comments 👇
I'll reveal the correct answer and the why tomorrow.
1
5 comments
06/23 Question of the Day
skool.com/anthropic-claude-architects-7929
The home for Claude Architects, aspiring architects — and everyone building with Claude.
Leaderboard (30-day)
1
+47
2
+47
3
+39
4
+22
5
+14
Powered by