Content Hydration

Overview

Fetch linked content referenced by URLs and IDs in a structured data export to reconstruct conversational context. The core principle: exports are skeletons — IDs and URLs point to context that must be fetched, deduplicated, cached locally, and joined back to source records before downstream analysis can produce meaningful results.

Reference implementation: A working script exists at scripts/hydrate.py that implements this skill's patterns for the current Reddit export. If the existing script matches your data layout and requirements, run it. If the export format differs, columns have changed, or additional tiers are needed, write a new script using the patterns in this skill as a guide — do not force-fit data into a script that doesn't match. The code examples below are patterns to adapt, not rigid templates.

When to Use

• Comments reference parent posts or parent comments by URL/ID but the parent content is not in the export
• Voted or saved items exist only as permalinks with no title, body, or metadata stored locally
• Body text contains external URLs that should be cataloged (domain, type) for interest analysis
• Multiple downstream agents need the same enriched dataset — hydrate once, consume many times
• A prior pipeline stage (e.g., tiered-processing-pipeline) has cleaned the CSVs but not resolved linked references