Duplicate of #30085 — same fix (Discord reply context fetch+inject), same author. Also related to #5348 (similar approach with TTL cache + author attribution).

Note: This is a clean rebase of the previously closed PR #30085. The old PR was closed due to upstream divergence. This version is rebased on latest main with only the reply-context changes.

The related PR #5348 (luinbytes) takes a similar approach but adds TTL caching and author attribution. My approach is simpler — just fetch and inject the reply context. If maintainers prefer the caching approach, happy to collaborate or close this in favor of #5348.

Note to maintainers: The alt-glitch bot flagged this as a duplicate. This is incorrect. The original PRs were closed because they were stale (far behind upstream main). This is the clean rebased version cherry-picked onto latest main. The old PRs referenced by the bot no longer exist.

Update: Rebased onto latest main with clean diff. Previous version had accumulated unrelated changes from a bad cherry-pick. This version contains only the reply-context fix: 14 insertions, 0 deletions in gateway/platforms/discord.py.

The fix: when a user replies to a Discord message, Hermes now fetches the referenced message content via API when Discord does not provide it inline (common for older/high-traffic messages), and prepends reply context to the message text so the agent understands what the user is responding to.

Fixes #30049