ironbee import backfills them into the IronBee Collector so your analytics include history, not just sessions from today forward. Discovery scans all three stores (~/.claude/projects/, ~/.codex/sessions/, and Cursor’s local state.vscdb) and routes each session through the right importer automatically, with no client flag needed.
Sessions already tracked are skipped automatically, so the command is safe to re-run.
Basic usage
From a project directory, import that project’s history:Cursor has no per-session transcript file — its “transcript” is the composer stored in
state.vscdb. Token and cost figures for imported Cursor sessions come from Cursor’s usage API and require cursor.apiAccess to be on (the default); without it, imported Cursor sessions carry structure only.Choosing what to import
Pick at most one scope (the default is the current directory):Limiting the time range
Pick at most one (the default is no time filter):Preview and safety
Always start with a dry run if you’re unsure of the volume:
A permanent rejection from the Collector (bad credential, exhausted quota, malformed batch — any non-
429 4xx) aborts the import immediately with the HTTP status, instead of burning through the retry backoff. Only transient failures (network, 429, 5xx) are retried.Common scenarios
What’s next?
Inspecting sessions
Check verdicts and session state from the terminal.
Open the Console
Your imported sessions appear alongside live ones, with the same analytics and timelines.