New Peptide_Linker class scans generated HTML and wraps every mention
of a known peptide in a hyperlink to its /peptides/ database page.
Runs as post-processing in Publisher — no LLM involvement, 100% reliable.
Handles case-insensitive matching, hyphen/space variants, skips existing
<a> tags. Gracefully no-ops if PR Core is not active.
Removed prompt-based peptide linking from Content_Prompts (unreliable).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When any peptide from the PR Core database is mentioned in a generated
article, the first mention is now automatically linked to its /peptides/
page. Gracefully degrades if PR Core is not active.
Extracted all prompt construction into Content_Prompts (252 lines),
slimming Content_Generator from 324 → 162 lines. Both under 300.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
