Completed background/remote-agent task output can't be re-fetched once the task ID expires

Summary

Once a background task (e.g. a remote_agent task like /ultrareview) completes and its result is delivered via <task-notification>, there appears to be no way to retrieve that exact raw result again later — even in the same session. The task ID stops resolving shortly after completion, and there's no stable link/output file that survives independently.

Repro

1. Ran /ultrareview on a branch. It launched a cloud review and returned a task ID plus a tracking URL (https://claude.ai/code/session_...).
2. The review completed; a <task-notification> delivered the full JSON findings payload inline in the conversation.
3. Visited the tracking URL afterward — no output was shown there.
4. Later in the same session, called TaskOutput with the same task ID to try to re-fetch the raw payload:

``
TaskOutput({ task_id: "rccpjvyqu", block: false })
→ Error: No task found with ID: rccpjvyqu
``

Expected

Some durable way to get back the exact raw output of a completed task — either:

• the tracking URL actually shows the completed session's output, or
• TaskOutput (or an equivalent) can still retrieve a completed task's result for some reasonable retention window, or
• the task notification includes a stable output file path (as it does for local bash/agent tasks) that persists on disk.

Actual

• The tracking URL showed nothing.
• The task ID was already unresolvable via TaskOutput well within the same conversation/session.
• The only "copy" of the results left is what the assistant had already reproduced in its own chat output (i.e. regenerated through the model), which the user reasonably doesn't consider an exact-fidelity source.

Impact

Users can't get a guaranteed byte-exact copy of an expensive cloud review's findings after the fact — they have to trust the assistant's transcription of the JSON, or lose access to the original entirely once the task ID expires.