i was wondered what take so much memory and made some tool over cursor-chat-browser to visualize that - GitHub - dvv47/cursor-chat-browser: A web application for browsing and managing chat histories from the Cursor editor's AI chat feature. View, search, and export your AI conversations in various formats.
You can try it:
so as i understood this - it is a history for some tool calls
for example one of my 14mb database entity is web_search tool call that downloaded some pdf inside db:
bubbleId:fde1b51f-cc0c-46f3-a0ca-977d24ad6843:d308f156-7be7-40b9-8c8f-dff65eee3520
"toolFormerData": {
"tool": 18,
"toolCallId": "toolu_vrtx_01NVMJs6uUKSwkQBti4K7wkv",
"status": "completed",
"rawArgs": "{\"search_term\": \"PyMesh2 vs PyMesh form_mesh alternative\", \"explanation\": \"Looking for information about the differences between PyMesh versions and how to replace form_mesh\"}",
"name": "web_search",
"params": "{\"searchTerm\":\"PyMesh2 vs PyMesh form_mesh alternative\"}",
"additionalData": {},
"userDecision": "accepted",
"result": "{\"references\":[{\"title\":\"PyMesh Documentation - Read the Docs\",\"url\":\"https://media.readthedocs.org/pdf/pymesh/latest/pymesh.pdf\",\"chunk\":\"%PDF-1.5 %����...
another one is incredable size of json 15mb of text related to some code analysis:
"hasChangedContext": true,
"capabilities": [
{
"type": 15,
"data": {
"bubbleDataMap": "{\"767e9484-7ebf-4a07-946b-78a605993ee3\":{\"tool\":9,\"toolCallId\":\"toolu_vrtx_01GwESxqfXz9Shwwy8cDKAP5\",\"status\":\"completed\",\"rawArgs\":\"{\\\"query\\\": \\\"celery_app.py initialize_all tasks\\\", \\\"target_directories\\\": [\\\"error_analyzer_backend\\\"]}\",\"name\":\"codebase_search\",\"params\":\"{\\\"repositoryInfo\\\":{\\\"relativeWorkspacePath\\\":\\\".\\\",\\\"repoName\\\":\\\"5b71baf7-1f7f-41ff-a98a-754dc4b767d1\\\",\\\"repoOwner\\\":\\\"google-oauth2|user_01JQ9PZB83102BYE0XVE6S179M\\\",\\\"orthogonalTransformSeed\\\":1289528086572099,\\\"preferredEmbeddingModel\\\":\\\"EMBEDDING_MODEL_UNSPECIFIED\\\",\\\"workspaceUri\\\":\\\"[object Object]\\\"},\\\"query\\\":\\\"celery_app.py initialize_all tasks\\\",
composerData:93b241dc-f4f0-43d4-91f1-11ca7d798df5
Probably all of this is fine. But as any time cursour trying to work with this database with direct i/o operations to file its freezing a lot. So will be great to work with this database in RAM
And may be have some tools to clear old chat history