Feature request: continuously ingest GitHub releases for top packages to reduce stale knowledge

Problem

Claude's biggest practical gap is recency. Training data is a static snapshot, so Claude confidently suggests deprecated APIs, outdated defaults, and patterns that were correct 6 months ago but aren't anymore.

Proposed solution

Continuously ingest release notes and changelogs from the top N most-depended-on packages (e.g., top 10,000 by dependents across npm, PyPI, crates.io, Maven, etc.) via the GitHub Releases API. Fold this into Claude's available knowledge — either at training time or as a retrieval layer.

The data is already structured and available (GET /repos/{owner}/{repo}/releases). The value is high: release notes are curated, authoritative summaries of what changed and why — far better signal-to-noise than raw commits.

Why this matters

When Claude helps with code, the most common failure mode is stale knowledge — wrong versions, renamed APIs, deprecated patterns. This single improvement would close the largest gap in practical usefulness for software engineering tasks.