Pre-compaction conversation is unreadable from /resume — users must hand-parse session .jsonl files

Problem

After a long session compacts (auto-compaction on context exhaustion, or manual /compact), there is no way in the product to read the earlier, pre-compaction part of the conversation. /resume only reopens the live (compacted) session; the original messages are not browsable anywhere in the UI, even though they are all still stored in the session's .jsonl under ~/.claude/projects/<project>/.

Today I needed to re-read a multi-day design discussion from before two compaction points. The only way was to locate the right .jsonl by file size and mtime, then parse JSON Lines by hand (grep/jq/python) to recover my own conversation. That is a terrible user experience for something the product already has on disk in full.

Request

Make the full transcript, including pre-compaction history, readable from the product itself, e.g.:

• /resume (or the session picker) offering a read-only view of the complete transcript, with compaction points marked; and/or
• a /transcript or claude export command that renders the session .jsonl to readable markdown/HTML.

Environment

• Claude Code on Windows 11 (desktop app + CLI), background job session
• Session file confirmed to contain the entire history across both compaction points (1,672 entries over 3 days), so this is purely a viewing/UX gap, not data loss

🤖 Filed with Claude Code on the user's behalf