File size: 16,761 Bytes
4e57bf4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 |
# =============================================================================
# Phishing Campaign Setup Assistant
# =============================================================================
# Description: A Gradio-based chatbot application using LangChain and OpenAI
# to guide users through setting up a phishing simulation campaign step-by-step.
#
# Requirements:
# - Python 3.x
# - Libraries: langchain, langchain_openai, langchain_community, gradio,
# python-dotenv, google-generativeai
# - Environment Variables (.env file):
# - OPENAI_API_KEY
# - GOOGLE_API_KEY
# - Data Files (in the same directory):
# - company_info.json
# - user_info.json
# =============================================================================
# --- 0. Required Imports ---
# Standard library imports
import os
import datetime
import json
import re
import base64
import tempfile
# Third-party imports for AI & LLMs
from dotenv import load_dotenv
from openai import OpenAI
from google import genai as google_genai
from google.genai import types as google_genai_types
from langchain.agents import create_openai_tools_agent, AgentExecutor
from langchain_openai import ChatOpenAI
from langchain_core.tools import StructuredTool
from langchain_core.messages import HumanMessage, AIMessage
from langchain import hub
from langchain_community.tools import DuckDuckGoSearchRun
# Third-party import for Web UI
import gradio as gr
# --- 1. Configuration and Initialization ---
# Load environment variables from a .env file
load_dotenv()
# Initialize the OpenAI client for the LangChain agent
# We use a low temperature (0.0) for predictable, task-oriented behavior.
llm = ChatOpenAI(model="gpt-4o", temperature=0.0)
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
# Initialize the Google GenAI Client for the image generation tool
# google_genai.configure(api_key=os.getenv("GOOGLE_API_KEY"))
genai_client = google_genai.Client(api_key=os.getenv("GOOGLE_API_KEY"))
# --- 2. Tool Definitions ---
# These functions define the actions (tools) the AI agent can perform.
def generate_image(prompt: str) -> dict:
"""
Generates an image based on a text prompt, saves it to 'generated_phishing_image.png'
in the current directory (overwriting previous images), and returns the absolute file path.
"""
# Fixed filename ensures replacement on subsequent generations.
output_filename = "generated_phishing_image.png"
print(f"INFO: Generating image with prompt: '{prompt}'")
try:
output = genai_client.models.generate_images(
prompt=prompt,
model="imagen-4.0-generate-preview-06-06",
config=google_genai_types.GenerateImagesConfig(
number_of_images=1,
aspect_ratio="16:9",
),
)
generated_img = output.generated_images[0].image
# Save the image to the fixed path in the current directory.
generated_img.save(output_filename)
# Get the absolute path for reliable referencing in the HTML.
absolute_image_path = os.path.abspath(output_filename)
print(f"INFO: Image saved to: {absolute_image_path}")
return {"status": "success", "image_path": absolute_image_path}
except Exception as e:
print(f"ERROR: Image generation failed: {e}")
return {"status": "error", "message": f"Image generation failed: {e}"}
def get_company_info() -> dict:
"""
Retrieves company information (name, logoUrl, departments, etc.) from company_info.json.
"""
print("INFO: Reading company_info.json")
try:
with open('company_info.json', 'r') as f:
data = json.load(f)
return {"status": "success", "data": data}
except FileNotFoundError:
return {"status": "error", "message": "company_info.json not found."}
except json.JSONDecodeError:
return {"status": "error", "message": "Error decoding company_info.json."}
def get_user_info() -> dict:
"""
Retrieves the current user's information (name, role, email) from user_info.json.
"""
print("INFO: Reading user_info.json")
try:
with open('user_info.json', 'r') as f:
data = json.load(f)
return {"status": "success", "data": data}
except FileNotFoundError:
return {"status": "error", "message": "user_info.json not found."}
except json.JSONDecodeError:
return {"status": "error", "message": "Error decoding user_info.json."}
def create_html_template(html_code: str) -> dict:
"""
Takes a complete HTML string, cleans it (removes newlines), and prepares it for preview.
"""
print("INFO: Formalizing agent-generated HTML template.")
# Clean HTML by removing newlines for compact storage/transmission
cleaned_html = html_code.replace("\n", "").replace("\r", "")
return {"status": "success", "template": cleaned_html}
def send_test_email(recipient: str, html_body: str) -> dict:
"""Simulates sending a test phishing email to a specified recipient."""
print(f"INFO: Test email sent to {recipient}")
return {"status": "success", "data": {"recipient": recipient}, "message": f"Test email sent to {recipient}."}
def get_or_create_employee_list(action: str, employee_data: list = None) -> dict:
"""Simulates managing employee lists (create, add, use_existing)."""
message = f"Action '{action}' on employee list was successful."
return {"status": "success", "data": {"action": action}, "message": message}
def select_target_group(group_type: str, values: list = None) -> dict:
"""
Selects the target group (all, department, individual). Includes error checking
to ensure 'values' are provided when necessary.
"""
if group_type == "all":
message = "The campaign will target all employees."
elif group_type == "department" and values:
message = f"Targeting departments: {', '.join(values)}."
elif group_type == "individual" and values:
message = f"Targeting individuals: {', '.join(values)}."
else:
# Handle cases where 'values' are missing or the group_type is unknown.
message = f"Error: Invalid selection for group type '{group_type}' or missing values."
return {"status": "success", "data": {"group_type": group_type, "targets": values}, "message": message}
def schedule_attack(date_time: str) -> dict:
"""Simulates scheduling the phishing campaign."""
return {"status": "success", "data": {"scheduled_for": date_time},
"message": f"Campaign scheduled for {date_time}."}
# --- 3. Agent and Prompt Configuration ---
# Assemble all functions into a list of StructuredTools for the agent
tools = [
StructuredTool.from_function(func=generate_image, name="GenerateImage",
description="Generates an image from a prompt and returns its local file path."),
StructuredTool.from_function(func=get_company_info, name="GetCompanyInfo",
description="Retrieves company information (including logoUrl and departments)."),
StructuredTool.from_function(func=get_user_info, name="GetUserInfo",
description="Retrieves the current user's information (including email)."),
StructuredTool.from_function(func=create_html_template, name="CreateHtmlTemplate",
description="Finalizes the phishing email's HTML code."),
StructuredTool.from_function(func=send_test_email, name="SendTestEmail",
description="Sends a test phishing email for review."),
StructuredTool.from_function(func=get_or_create_employee_list, name="ManageEmployeeList",
description="Manages the employee list for the campaign."),
StructuredTool.from_function(func=select_target_group, name="SelectTargetGroup",
description="Selects the target group for the campaign."),
StructuredTool.from_function(func=schedule_attack, name="ScheduleAttack",
description="Schedules the phishing campaign.")
]
# Pull a standard agent prompt template from the LangChain hub
prompt = hub.pull("hwchase17/openai-tools-agent")
# Define the master instructions for the AI agent (the "System Prompt")
SYSTEM_PROMPT = """
You are an AI assistant named Cbulwork, designed to set up phishing simulation campaigns. Your goal is to guide the user step-by-step with precision and clarity. The user has already been greeted, so you should start directly with the process.
**PROCESS:**
**Step 1: Gather Context & Suggest Scenario**
- Call `GetUserInfo` and `GetCompanyInfo`.
- Greet the user by name.
- If the user has NOT provided a topic, suggest 5 relevant scenarios based on company info.
- Await the user's confirmation of the scenario.
**Step 2: Choose Template Type**
- Ask the user to choose a template type: Text Only, Text + Photo, or Photo Only.
- Wait for their selection.
**Step 3: Template Design**
- Write a **highly detailed and convincing**, valid HTML code for the email based on the user's choice.
- **IMAGE & LOGO RULES (CRITICAL):**
- If 'Text + Photo' or 'Photo Only' was chosen:
1. Call `GenerateImage`. The prompt MUST be for a **flyer-style image with simple, bold text** related to the scenario (e.g., "A modern corporate flyer with the text 'Urgent Action Required: Update Your Password'").
2. Use the exact `image_path` returned by the tool in the `src` attribute of an `<img>` tag. **You MUST prefix the local path with `file:///` for the preview to work.**
- If "Text + Photo" was chosen, also include the `logoUrl` from `GetCompanyInfo` in a separate `<img>` tag.
- **CONTENT RULES:**
- The email body must have at least two convincing paragraphs.
- Generate a professional footer with fake details (address, contact info) for realism.
- Generate a compelling subject, personalized greeting ("{{recipient.name}}"), detailed body, footer, and a clear call-to-action.
- Do NOT include copyright lines.
- After writing the code, you MUST call `CreateHtmlTemplate` with the HTML as a single string.
**Step 4: Send Test Email**
- After approval, ask to send a test email. If yes, use `SendTestEmail` with the user's email.
**Step 5: Employee List**
- Ask for the list provision method (upload/manual). If manual, provide an example format (`Name,Email`). Call `ManageEmployeeList`.
**Step 6: Target Group Selection**
- Ask to target 'all', 'department', or 'individual'.
- If not 'all', ask for the specific names/departments (list available departments from `GetCompanyInfo`).
- Call `SelectTargetGroup` with the correct `group_type` and `values`.
**Step 7: Schedule Campaign**
- Ask for a future launch date/time (`dd/mm/yyyy` format). Call `ScheduleAttack`.
**Step 8: Final Summary & Confirmation**
- Provide a complete summary. Ask for final confirmation. After confirmation, ask if there is anything else.
"""
# Insert the system prompt into the template
prompt.messages[0].prompt.template = SYSTEM_PROMPT
# Create the agent (LLM + Tools + Prompt)
agent = create_openai_tools_agent(llm, tools, prompt)
# Create the agent executor (the runtime for the agent)
agent_executor = AgentExecutor(
agent=agent,
tools=tools,
verbose=True, # Set to True to see the agent's thought process and tool usage in the console
handle_parsing_errors=True,
max_iterations=15,
return_intermediate_steps=True # Required to capture tool output for the UI
)
# --- 4. Core Application Logic ---
def run_agent_turn(user_input: str, chat_history: list) -> dict:
"""
Processes one turn of the conversation: sends input to the agent, executes tools,
and collects the results (response, HTML, image path, and tool calls).
"""
# Convert Gradio chat history format to LangChain message format
langchain_messages = [
HumanMessage(content=msg["content"]) if msg["role"] == "user" else AIMessage(content=msg["content"])
for msg in chat_history
]
# Invoke the agent
response = agent_executor.invoke({
"input": user_input,
"chat_history": langchain_messages
})
agent_output = response.get("output", "Sorry, an error occurred.")
# Initialize variables to capture outputs from the agent's steps
html_to_preview = ""
generated_image_path = None
function_calls = []
intermediate_steps = response.get("intermediate_steps", [])
# Process the steps the agent took
for action, tool_output in intermediate_steps:
# Log the tool call for the JSON output box
function_calls.append({
"tool_name": action.tool,
"tool_args": action.tool_input,
"tool_output": tool_output,
})
# Capture the HTML output if the CreateHtmlTemplate tool was used
if action.tool == "CreateHtmlTemplate" and isinstance(tool_output, dict):
html_to_preview = tool_output.get("template", "")
# Capture the image path if the GenerateImage tool was used successfully
if action.tool == "GenerateImage" and tool_output.get("status") == "success":
generated_image_path = tool_output.get("image_path")
# Update the chat history
updated_chat_history = chat_history + [
{"role": "user", "content": user_input},
{"role": "assistant", "content": agent_output}
]
# Return a structured dictionary with all results
return {
"agent_response": agent_output,
"html_preview": html_to_preview,
"function_calls": function_calls,
"updated_chat_history": updated_chat_history,
"generated_image_preview": generated_image_path
}
def process_input_for_gradio(user_input: str, chat_history: list) -> tuple:
"""
Event handler for the Gradio UI. Calls the core agent logic and returns
the outputs in the order expected by the Gradio outputs list.
"""
if not user_input.strip():
# Don't process empty input
return chat_history, "", None, None
# Run the agent turn
json_output = run_agent_turn(user_input, chat_history)
# Optional: Print the backend output to the console for debugging
print(f"--- Backend JSON Output ---\n{json.dumps(json_output, indent=2)}\n--------------------------")
# Return the data in the order of the Gradio outputs=[...] list
return (
json_output["updated_chat_history"],
json_output["html_preview"],
json_output["function_calls"],
json_output["generated_image_preview"]
)
# --- 5. Gradio User Interface Definition ---
# Define the UI layout using Gradio Blocks
with gr.Blocks(theme=gr.themes.Default(primary_hue="blue", secondary_hue="sky")) as demo:
gr.Markdown("## Phishing Campaign Setup Assistant")
gr.Markdown("I will guide you step-by-step to create and schedule a new phishing campaign.")
with gr.Row():
# Left Column: Chat Interface
with gr.Column(scale=1):
welcome_message = "Hello, I'm your AI phishing assistant. Send a message to get started."
chatbot = gr.Chatbot(
value=[{"role": "assistant", "content": welcome_message}],
label="Conversation",
height=600,
type="messages" # Ensures we use the modern {'role': '...', 'content': '...'} format
)
user_input = gr.Textbox(
placeholder="Send a message to continue...",
label="Your Message",
scale=12
)
# Right Column: Previews and Debugging
with gr.Column(scale=1):
gr.Markdown("### Email Template Preview")
html_block = gr.HTML(label="HTML Preview")
gr.Markdown("### Generated Image Preview")
# Added an Image component to display the generated flyer/image
image_preview_box = gr.Image(label="Image Preview", interactive=False)
gr.Markdown("### Function Call Output (Debugging)")
json_requests_box = gr.JSON(label="Function 'Requests' Output")
# Connect the user input submission to the event handler
user_input.submit(
fn=process_input_for_gradio,
inputs=[user_input, chatbot],
# Ensure outputs match the return tuple of process_input_for_gradio
outputs=[chatbot, html_block, json_requests_box, image_preview_box]
)
# Clear the input box after submission
user_input.submit(lambda: "", None, user_input)
# --- 6. Application Launch ---
if __name__ == "__main__":
# Launch the Gradio web server
print("Launching Phishing Campaign Setup Assistant UI...")
demo.launch(debug=False) |