Skip to content

Internal Tools in Conversation

Overview

Internal tools provide built-in capabilities that enhance your conversations and agents without requiring external integrations. Currently, Elitea supports the Python Sandbox internal tool, which enables secure Python code execution directly within your conversations using Pyodide (Python compiled to WebAssembly).

Use cases:

  • Execute Python code snippets during conversations
  • Perform calculations and data analysis
  • Test Python algorithms in a secure environment
  • Process data with Python libraries
  • Generate visualizations and reports

Prerequisites

  • Permission Level: User role with conversation edit access
  • Conversation: An active conversation or agent configuration
  • Backend: Deno must be installed on the server (handled by system administrators)

Enabling Internal Tools in Conversations

Step 1: Open the Internal Tools Configuration

  1. Navigate to your conversation
  2. Locate the chat input toolbar at the bottom of the screen
  3. Find the Internal Tools icon (value icon) next to the attachment button
  4. Click the Internal Tools icon to open the configuration popup

Step 2: Enable Python Sandbox

  1. In the configuration popup, you'll see a list of available internal tools
  2. Find Python sandbox in the list
  3. Click the toggle switch next to "Python sandbox" to enable it
  4. The tool description reads: "Enable Python code execution in a secure sandbox using Pyodide"

Enable Python Sandbox

Step 3: Verify Enablement

  1. Once enabled, the toggle switch will move to the ON position
  2. A success toast notification will appear: "Internal tools configuration updated"
  3. The configuration is saved immediately to the conversation metadata
  4. Click anywhere outside the popup to close it

Enabling Internal Tools in Agent Configuration

You can also configure internal tools as part of an agent's default configuration.

Step 1: Access Agent Configuration

  1. Navigate to Agents in the main menu
  2. Select the agent you want to configure
  3. Click the Configuration tab
  4. Scroll to the TOOLKITS section

Step 2: Enable Python Sandbox for the Agent

  1. In the TOOLKITS section, find the Python sandbox switch at the bottom
  2. Toggle the switch to enable Python sandbox for this agent
  3. The setting includes an info icon (ℹ️) with a tooltip explaining the feature
  4. This configuration is saved in the agent version metadata

Step 3: Save Changes

  1. Click Save at the top of the configuration page
  2. The Python sandbox will now be available in all new conversations using this agent
  3. Existing conversations are not affected and maintain their own internal tools settings

Agent Python Sandbox

Using the Python Sandbox

Once enabled, the AI assistant can execute Python code during conversations.

Example conversation:

User: I'm writing a function to validate email addresses. Can you help me test it with various inputs to make sure it works correctly?

import re

def validate_email(email):
    """
    Improved email validation with better edge case handling.
    """
    # Check for consecutive dots
    if '..' in email:
        return False

    # More strict regex pattern
    # Local part: starts and ends with alphanumeric, can contain ._+-% 
    # Domain: proper format with labels that start/end with alphanumeric
    pattern = (
        r'^[a-zA-Z0-9][a-zA-Z0-9._%+-]*[a-zA-Z0-9]'
        r'@[a-zA-Z0-9]+([.-][a-zA-Z0-9]+)*\.[a-zA-Z]{2,}$'
        r'|^[a-zA-Z0-9]@[a-zA-Z0-9]+([.-][a-zA-Z0-9]+)*\.[a-zA-Z]{2,}$'
    )

    return re.match(pattern, email) is not None

Assistant: I'll help you test your email validation function with a comprehensive set of test cases! Let me run various inputs through it to identify any edge cases.

[Tool Used: pyodide_sandbox]

import re

def validate_email(email):
    """
    Improved email validation with better edge case handling.
    """
    # Check for consecutive dots
    if '..' in email:
        return False

    # More strict regex pattern
    pattern = r'^[a-zA-Z0-9][a-zA-Z0-9._%+-]*[a-zA-Z0-9]@[a-zA-Z0-9]+([.-][a-zA-Z0-9]+)*\.[a-zA-Z]{2,}$|^[a-zA-Z0-9]@[a-zA-Z0-9]+([.-][a-zA-Z0-9]+)*\.[a-zA-Z]{2,}$'

    return re.match(pattern, email) is not None

# Test cases
test_cases = [
    ("user@example.com", True),
    ("user.name@example.co.uk", True),
    ("user+tag@example.com", True),
    ("invalid.email@", False),
    ("@example.com", False),
    ("user@.com", False),
    ("user@example", False),
    ("user name@example.com", False),
    ("user@example..com", False),
]

print("Email Validation Test Results")
print("=" * 50)
passed = 0
failed = 0

for email, expected in test_cases:
    result = validate_email(email)
    status = "✓ PASS" if result == expected else "✗ FAIL"

    if result == expected:
        passed += 1
    else:
        failed += 1

    print(f"{status} | {email:30} | Expected: {expected}, Got: {result}")

print("=" * 50)
print(f"Results: {passed}/{len(test_cases)} passed, {failed} failed")

if failed > 0:
    print("\n⚠️  Some test cases failed. Review the pattern.")
else:
    print("\n✓ All test cases passed!")

Output: 

Email Validation Test Results
==================================================
✓ PASS | user@example.com              | Expected: True, Got: True
✓ PASS | user.name@example.co.uk       | Expected: True, Got: True
✓ PASS | user+tag@example.com          | Expected: True, Got: True
✓ PASS | invalid.email@                | Expected: False, Got: False
✓ PASS | @example.com                  | Expected: False, Got: False
✓ PASS | user@.com                     | Expected: False, Got: False
✓ PASS | user@example                  | Expected: False, Got: False
✓ PASS | user name@example.com         | Expected: False, Got: False
✓ PASS | user@example..com             | Expected: False, Got: False
==================================================
Results: 9/9 passed, 0 failed

✓ All test cases passed!

Great! Your improved email validation function passes all test cases. The enhancements you made successfully handle: - Consecutive dots (rejected) - Domain without TLD (rejected) - Proper multi-part domains (accepted) - Special characters in local part (accepted when appropriate)

Use Python Sandbox

The assistant can:

  • Install Python packages via Pyodide (e.g., numpy, pandas, matplotlib)
  • Execute multi-line Python scripts
  • Maintain state between code executions in the same conversation
  • Generate data visualizations
  • Process files and data

Verification Steps

For Conversations

  1. Check the icon state: The Internal Tools icon should be highlighted when tools are configured
  2. Test execution: Ask the assistant to execute a simple Python calculation

For Agents

  1. Review agent configuration: Check the Configuration tab to see enabled internal tools
  2. Start new conversation: Create a new conversation with the agent
  3. Verify availability: Confirm the Python sandbox is available without manual enablement

Effects & Caveats

What Happens When You Enable Python Sandbox

Immediate effects:

  • The AI assistant gains the ability to execute Python code
  • Code runs in a secure, isolated WebAssembly environment
  • No access to your local file system or network (unless explicitly configured server-side)
  • Execution results are returned in the conversation

Performance considerations:

  • First code execution may take longer due to Pyodide initialization
  • Subsequent executions are faster due to caching
  • Complex computations or large data processing may have timeout limits

Important Caveats

Security & Isolation

While the Python sandbox is isolated using Pyodide/WebAssembly, you should still review any code the assistant plans to execute, especially when working with sensitive data.

Package Availability

Not all Python packages are available in Pyodide. See the Pyodide packages list for supported packages.

Stateful Execution

By default, the sandbox maintains state between executions within the same conversation. Variables and imports persist across multiple code blocks.

Limitations

  • File I/O: Limited file system access (no access to local files)
  • Network: Restricted network access depending on server configuration
  • System Operations: Cannot execute system-level commands
  • Packages: Limited to Pyodide-compatible packages
  • Execution Time: Long-running scripts may timeout

Troubleshooting

Issue: Internal Tools Icon Not Visible

Possible causes:

  • You're on the Agents page (internal tools config is only available in conversations)
  • The conversation is loading
  • You're in a streaming response

Solution:

  • Wait for the page to fully load
  • Wait for any streaming responses to complete
  • Ensure you're in a conversation view, not the Agents list

Issue: "Failed to update internal tools config" Error

Possible causes:

  • Network connectivity issues
  • Backend API unavailable
  • Insufficient permissions

Solution:

  • Check your internet connection
  • Refresh the page and try again
  • Contact your administrator if the issue persists

Issue: Python Code Fails to Execute

Possible causes:

  • Deno not installed on the server
  • Package not available in Pyodide
  • Syntax errors in the code
  • Timeout exceeded

Solution:

  • Contact system administrator about Deno installation
  • Use Pyodide-compatible packages
  • Review error messages in the assistant's response
  • Break large computations into smaller chunks

Issue: Configuration Not Saving in Agent

Possible causes:

  • Form validation errors
  • Missing required fields
  • Conflicting configuration

Solution:

  • Ensure all required agent fields are filled
  • Save the entire configuration form, not just the internal tools toggle
  • Check for any error messages at the top of the form

Related Documentation