Cursor SQLite Data Extractor: Windows Conversations and Prompts. Python scripts to scan SQLite database files (`state.vscdb` and `state.vscdb.backup`) used by Cursor for extracting conversation history and text-based prompts. If conversation data is not available, the scripts fall back to extracting prompts (e.g., `textDescription` fields). Outp…

extract_conversations_prompts.py

#!/usr/bin/env python3
# wsl

import sqlite3
import json
from pathlib import Path
from datetime import datetime

# path to WorkspaceStorage on Windows or any paths with folders and state.vscdb inside 
ENTRY_PATH="/mnt/c/Users/manoj/AppData/Roaming/Cursor/User/workspaceStorage"

def extract_conversations(cursor, table_name):
    """
    Try to extract conversation data from the specified table.
    Looks for dicts containing 'conversation' fields.
    """
    print(f"  Searching for conversations in table '{table_name}'...")
    conversations = []
    try:
        cursor.execute(f"SELECT key, value FROM {table_name} LIMIT 1000;")
        rows = cursor.fetchall()
        for key, value in rows:
            try:
                # Decode value if it's a BLOB
                if isinstance(value, bytes):
                    value = value.decode('utf-8')

                # Attempt to parse JSON
                parsed = json.loads(value) if value else None

                # Check for conversation-like data
                if isinstance(parsed, dict) and 'conversation' in parsed:
                    conversations.append({'key': key, 'conversation': parsed['conversation']})
            except:
                # Ignore malformed rows
                pass
    except Exception as e:
        print(f"  Error extracting conversations from '{table_name}': {e}")
    return conversations

def extract_prompts(cursor, table_name):
    """
    Try to extract prompts using the 'textDescription' field or related data from the specified table.
    """
    print(f"  Searching for prompts in table '{table_name}'...")
    prompts = []
    try:
        cursor.execute(f"SELECT key, value FROM {table_name} LIMIT 1000;")
        rows = cursor.fetchall()
        for key, value in rows:
            try:
                # Decode value if it's a BLOB
                if isinstance(value, bytes):
                    value = value.decode('utf-8')

                # Attempt to parse JSON
                parsed = json.loads(value) if value else None

                # Look for prompts in known fields
                if isinstance(parsed, list):
                    for item in parsed:
                        if isinstance(item, dict) and 'textDescription' in item:
                            prompts.append(item['textDescription'])
                elif isinstance(parsed, dict) and 'textDescription' in parsed:
                    prompts.append(parsed['textDescription'])
            except:
                # Ignore malformed rows
                pass
    except Exception as e:
        print(f"  Error extracting prompts from '{table_name}': {e}")
    return prompts

def scan_database(db_path):
    """
    Scan a single SQLite database for conversations and prompts.
    """
    print(f"\nScanning database: {db_path}")
    conversations = []
    prompts = []

    try:
        conn = sqlite3.connect(db_path)
        cursor = conn.cursor()

        # List all tables in the database
        cursor.execute("SELECT name FROM sqlite_master WHERE type='table';")
        tables = cursor.fetchall()

        for table_name, in tables:
            # Attempt to extract conversations
            conversations.extend(extract_conversations(cursor, table_name))
            # Attempt to extract prompts
            prompts.extend(extract_prompts(cursor, table_name))

        conn.close()
    except Exception as e:
        print(f"  Error scanning database {db_path}: {e}")

    return conversations, prompts

def save_to_markdown(output_file, conversations, prompts):
    """
    Save conversations and prompts to a Markdown file.
    """
    with open(output_file, 'a', encoding='utf-8') as f:
        if conversations:
            f.write("# Conversations\n")
            for conv in conversations:
                f.write(f"## Key: {conv['key']}\n")
                for message in conv['conversation']:
                    sender = "User" if message.get('type') == 1 else "Assistant"
                    text = message.get('text', '').strip()
                    f.write(f"**{sender}**: {text}\n")
                f.write("\n---\n")
        else:
            f.write("# No conversations found.\n")

        if prompts:
            f.write("\n# Prompts\n")
            for i, prompt in enumerate(prompts, 1):
                f.write(f"{i}. {prompt.strip()}\n")

def scan_directory(base_path):
    """
    Scan all SQLite databases in the directory and subdirectories for conversation and prompt data.
    """
    base_path = Path(base_path)
    print(f"Scanning directory: {base_path}")

    if not base_path.exists():
        print(f"Directory does not exist: {base_path}")
        return

    # Prepare an output Markdown file
    output_dir = Path("output")
    output_dir.mkdir(exist_ok=True)
    output_file = output_dir / f"conversations_and_prompts_{datetime.now().strftime('%Y%m%d_%H%M%S')}.md"

    db_files = list(base_path.rglob("state.vscdb")) + list(base_path.rglob("state.vscdb.backup"))
    if not db_files:
        print("No database files found.")
        return

    print(f"Found {len(db_files)} database files.")
    for db_file in db_files:
        conversations, prompts = scan_database(db_file)
        save_to_markdown(output_file, conversations, prompts)

    print(f"\nData extraction completed! Results saved to {output_file}")

def main():
    """
    Main function to scan a directory and extract conversations and prompts.
    """
    # Update this path to your workspaceStorage directory
    base_path = Path(ENTRY_PATH)
    scan_directory(base_path)

if __name__ == "__main__":
    main()


'''
Output looks like this:
(base) mbhat@manoj:~/cursor-stuff$ cat output/conversations_and_prompts_20250414_014537.md
# No conversations found.
# No conversations found.
# No conversations found.
# No conversations found.

# Prompts
1. write code to make a cybersecurity software
2. write code to make a cybersecurity software
3. write code to make a cybersecurity software
4. write code to make a cybersecurity software
5. write code for a simple cybersecurity software run using C++ backend
6. CMakeLists.txt
7. file_monitor.h
8. process_monitor.h
9. network_monitor.h
10. file_monitor.cpp
11. process_monitor.cpp
12. network_monitor.cpp
13. main.cpp
14. README.md
# No conversations found.

# Prompts
1. Create a website frontend that can use the repo as a service on a different server while the client requests it for 3D file from the input image on the frontend and the server creates it and sends the data to the client. I want to host it myself and be able to use it on mobile and any type of device to take pictures of objects around me from the camera of the system and use it to create 3D objects and save it as .fbx file or one that can be used on any convinent viewer. Also on the frontend there should be space to view the 3D file using threejs. Meaning from image to 3d object that the user can place it on a threejs scene
# No conversations found.
# No conversations found.
# No conversations found.

# Prompts
1. what is needed to do to install this on chrome ?
2. what is needed to do to install this on chrome ?
3. what is needed to do to install this on chrome ?
4. I need to modify the files so as to make it work with "Load unpacked" option
# No conversations found.
# No conversations found.
'''

scan_conversation_data.py

#!/usr/bin/env python3
#wsl

import sqlite3
import json
from pathlib import Path
import os

# path to WorkspaceStorage on Windows or any paths with folders and state.vscdb inside
ENTRY_PATH="/mnt/c/Users/manoj/AppData/Roaming/Cursor/User/workspaceStorage"

def inspect_table(cursor, table_name):
    """
    Inspect a table and search for any conversation-like data.
    Prints out potential matches.
    """
    print(f"Inspecting table: {table_name}")
    
    try:
        # Fetch a sample of rows
        cursor.execute(f"SELECT key, value FROM {table_name} LIMIT 100;")
        rows = cursor.fetchall()
        print(f"  Found {len(rows)} rows in table '{table_name}'.")
        
        for key, value in rows:
            # Attempt to decode the value
            try:
                if isinstance(value, bytes):
                    # Attempt to decode if the value is a BLOB
                    value = value.decode('utf-8')  
                
                # Try parsing as JSON
                parsed = json.loads(value) if isinstance(value, str) else None
                
                # Check for conversation-like data
                if isinstance(parsed, dict) and ('conversation' in parsed or 'messages' in parsed):
                    print(f"  Potential conversation found for key '{key}':")
                    print(json.dumps(parsed, indent=2))
                elif isinstance(parsed, list) and all(isinstance(item, dict) for item in parsed):
                    print(f"  Potential list of conversations/messages found for key '{key}':")
                    print(json.dumps(parsed[:2], indent=2))  # Show a sample of 2 items
            except Exception as e:
                # If decoding fails, just skip
                continue
    except Exception as e:
        print(f"  Error inspecting table '{table_name}': {e}")

def scan_database(db_path):
    """
    Scan a single SQLite database for tables and search for conversation data.
    """
    print(f"\nScanning database: {db_path}")
    try:
        conn = sqlite3.connect(db_path)
        cursor = conn.cursor()

        # List all tables in the database
        cursor.execute("SELECT name FROM sqlite_master WHERE type='table';")
        tables = cursor.fetchall()
        if not tables:
            print("  No tables found in the database.")
            return
        
        print(f"  Found tables: {[t[0] for t in tables]}")

        # Inspect each table for potential conversation data
        for table_name, in tables:
            inspect_table(cursor, table_name)
        
        conn.close()
    except Exception as e:
        print(f"  Error scanning database {db_path}: {e}")

def scan_directory(base_path):
    """
    Scan all `state.vscdb` and `state.vscdb.backup` files in a directory and subdirectories.
    """
    base_path = Path(base_path)
    print(f"Scanning directory: {base_path}")

    if not base_path.exists():
        print(f"Directory does not exist: {base_path}")
        return

    # Find all files named `state.vscdb` or `state.vscdb.backup`
    db_files = list(base_path.rglob("state.vscdb")) + list(base_path.rglob("state.vscdb.backup"))
    if not db_files:
        print(f"No database files found in {base_path}.")
        return

    print(f"Found {len(db_files)} database files.")
    for db_file in db_files:
        scan_database(db_file)

def main():
    """
    Main function to scan a directory for conversation-like data in SQLite databases.
    """
    # Update this path to your `workspaceStorage` directory
    base_path = Path(ENTRY_PATH)
    scan_directory(base_path)

if __name__ == "__main__":
    main()


'''
THe output is like this:

(base) mbhat@manoj:~/cursor-stuff$ python scan_conversation_data.py
Scanning directory: /mnt/c/Users/manoj/AppData/Roaming/Cursor/User/workspaceStorage
Found 17 database files.

Scanning database: /mnt/c/Users/manoj/AppData/Roaming/Cursor/User/workspaceStorage/0d3805e2acc52ab98013bbe30bcc4cdb/state.vscdb
  Found tables: ['ItemTable', 'cursorDiskKV']
Inspecting table: ItemTable
  Found 55 rows in table 'ItemTable'.
  Potential list of conversations/messages found for key 'workbench.auxiliarybar.viewContainersWorkspaceState':
[
  {
    "id": "workbench.panel.aichat",
    "visible": true
  },
  {
    "id": "workbench.panel.composerViewPane2",
    "visible": false
  }
]
  Potential list of conversations/messages found for key 'aiService.generations':
[]
  Potential list of conversations/messages found for key 'aiService.prompts':
[]
  Potential list of conversations/messages found for key 'history.entries':
[]
  Potential list of conversations/messages found for key 'interactive.sessions':
[]
  Potential list of conversations/messages found for key 'comments.continueOnComments':
[]
  Potential list of conversations/messages found for key 'chat.editing.autosaveDisabled':
[]
  Potential list of conversations/messages found for key 'workbench.panel.viewContainersWorkspaceState':
[
  {
    "id": "workbench.panel.markers",
    "visible": true
  },
  {
    "id": "workbench.panel.output",
    "visible": true
  }
]
  Potential list of conversations/messages found for key 'terminal.integrated.environmentVariableCollectionsV2':
[
  {
    "extensionIdentifier": "vscode.git",
    "collection": [
      [
        "GIT_ASKPASS",
        {
          "variable": "GIT_ASKPASS",
          "value": "c:\\Users\\manoj\\AppData\\Local\\Programs\\cursor\\resources\\app\\extensions\\git\\dist\\askpass.sh",
          "type": 1,
          "options": {
            "applyAtProcessCreation": true,
            "applyAtShellIntegration": false
          }
        }
      ],
      [
        "VSCODE_GIT_ASKPASS_NODE",
        {
          "variable": "VSCODE_GIT_ASKPASS_NODE",
          "value": "C:\\Users\\manoj\\AppData\\Local\\Programs\\cursor\\Cursor.exe",
          "type": 1,
          "options": {
            "applyAtProcessCreation": true,
            "applyAtShellIntegration": false
          }
        }
      ],
      [
        "VSCODE_GIT_ASKPASS_EXTRA_ARGS",
        {
          "variable": "VSCODE_GIT_ASKPASS_EXTRA_ARGS",
          "value": "",
          "type": 1,
          "options": {
            "applyAtProcessCreation": true,
            "applyAtShellIntegration": false
          }
        }
      ],
      [
        "VSCODE_GIT_ASKPASS_MAIN",
        {
          "variable": "VSCODE_GIT_ASKPASS_MAIN",
          "value": "c:\\Users\\manoj\\AppData\\Local\\Programs\\cursor\\resources\\app\\extensions\\git\\dist\\askpass-main.js",
          "type": 1,
          "options": {
            "applyAtProcessCreation": true,
            "applyAtShellIntegration": false
          }
        }
      ],
      [
        "VSCODE_GIT_IPC_HANDLE",
        {
          "variable": "VSCODE_GIT_IPC_HANDLE",
          "value": "\\\\.\\pipe\\vscode-git-d8efc8ddbb-sock",
          "type": 1,
          "options": {
            "applyAtProcessCreation": true,
            "applyAtShellIntegration": false
'''