How to get all chat history

[sarkar (Partha)]

Where chat history are stored? Hoe to get all chat history?

[deanrie]

Hi @sarkar

Maybe this helps:

[Guide][5 steps] Exporting chats & prompts from Cursor

Go to workspaceStorage. (On Windows, cd %APPDATA%\Cursor\User\workspaceStorage) You will see a list of folders whose names are md5 hashes. Open each of them. datasette state.vscdb Visit “http://localhost:8001/state?” in a browser and enter this query: SELECT rowid, [key], value FROM ItemTable WHERE [key] IN (‘aiService.prompts’, ‘workbench.panel.aichat.view.aichat.chatdata’) Parse the json based on your use case.

[sarkar (Partha)]

Thanks it’s worked for me

[deanrie]

I was glad to help.

[amapic (Amaury PICHAT)]

you rule

[SUNLAND]

another work around…
COPY and PASTE this instruction message and give to your Cursor agent to help you recover lost chat history:

"Please help me recover lost chat history from Cursor’s workspace storage. This is a two-step process:

1. First, create a Python script called export_chats.py that can search through SQLite databases in the workspace storage directory (on Mac: ~/Library/Application Support/Cursor/User/workspaceStorage , each containing a state.vscdb file). The script should search through chat data for specific time periods and keywords related to the lost content.
2. Then, create a Node.js script called clean_history.js that processes the exported chat history. This script should:
3. • Filter conversations based on project-specific keywords

• Clean up markdown formatting
• Remove irrelevant content
• Output cleaned history files for each team We’ll need to make multiple search passes with export_chats.py , refining the parameters each time to filter out noise and focus on relevant conversations. Once we have the raw chat data, we’ll use clean_history.js to organize and clean it. When we find matching content, help me reconstruct the code and implementation details from the chat fragments. Please guide me through this process step by step, starting with creating both scripts and then helping me search effectively by suggesting relevant search terms and filtering strategies based on what I’m trying to recover. Remember that chat history might contain code snippets, implementation discussions, and debugging sessions that can help piece together the lost work."
• Cursor will find and translate

[Ishkei]

Hope this helps

A comprehensive toolkit for recovering lost chat history from Cursor’s workspace storage

GitHub - Ishkei/cursor-chat-recovery: A comprehensive toolkit for recovering lost chat...

[kgal-akl (Kobbi Gal)]

I wrote a short script (using cursor ) to find the cursor workspace for a specific search term:

find_cursor_workspace() {
    local search_term=$1
    local base_dir="$HOME/Library/Application Support/Cursor/User/workspaceStorage"

    if [[ -z "$search_term" ]]; then
        echo "Usage: find_cursor_workspace <search_term>"
        return 1
    fi

    # 1. Find all state.vscdb files
    find "$base_dir" -name "state.vscdb" -type f | while read -r db_path; do
        # 2. Query the SQLite DB for the search term in the 'value' column
        # We use 'grep' on the output of the SQL query to check for the term
        local match=$(sqlite3 "$db_path" "SELECT value FROM ItemTable" 2>/dev/null | grep -i "$search_term")

        if [[ -n "$match" ]]; then
            local workspace_dir=$(dirname "$db_path")
            local workspace_json="$workspace_dir/workspace.json"

            if [[ -f "$workspace_json" ]]; then
                # 3. Extract the workspace name/path
                local ws_path=$(jq -r '.workspace // .folder' "$workspace_json")
                echo "-----------------------------------"
                echo "Match Found in: $(basename "$workspace_dir")"
                echo "Workspace:      $ws_path"
            fi
        fi
    done
}

hope this helps.