feat(bump version):

Merge branch 'main' of github.com:msoedov/agentic_security
fix(clean up):
2026-06-24 14:19:55 +02:00 · 2025-03-12 13:42:59 +02:00 · 2025-03-12 13:41:56 +02:00 · 2025-03-12 13:41:24 +02:00 · 2025-03-12 13:39:56 +02:00 · 2025-03-12 12:30:30 +01:00
53 changed files with 1355 additions and 1078 deletions
@@ -17,3 +17,4 @@ inv/
 scripts/
 docx/
 agentic_security.toml
+/venv
@@ -83,7 +83,7 @@ agentic_security --port=PORT --host=HOST

 ## UI 🧙

-<img width="100%" alt="booking-screen" src="https://res.cloudinary.com/dq0w2rtm9/image/upload/v1736433557/z0bsyzhsqlgcr3w4ovwp.gif">
+<img width="100%" alt="booking-screen" src="https://raw.githubusercontent.com/msoedov/agentic_security/refs/heads/main/docs/images/demo.gif">

 ## LLM kwargs

@@ -408,10 +408,15 @@ For more detailed information on how to use Agentic Security, including advanced

 ## Roadmap and Future Goals

- \[ \] Expand dataset variety
- \[ \] Introduce two new attack vectors
- \[ \] Develop initial attacker LLM
- \[ \] Complete integration of OWASP Top 10 classification
+
+
+We’re just getting started! Here’s what’s on the horizon:
+
+- **RL-Powered Attacks**: An attacker LLM trained with reinforcement learning to dynamically evolve jailbreaks and outsmart defenses.
+- **Massive Dataset Expansion**: Scaling to 100,000+ prompts across text, image, and audio modalities—curated for real-world threats.
+- **Daily Attack Updates**: Fresh attack vectors delivered daily, keeping your scans ahead of the curve.
+- **Community Modules**: A plug-and-play ecosystem where you can share and deploy custom probes, datasets, and integrations.
+

 | Tool                    | Source                                                                        | Integrated |
 |-------------------------|-------------------------------------------------------------------------------|------------|
@@ -439,4 +444,9 @@ Before contributing, please read the contributing guidelines.

 Agentic Security is released under the Apache License v2.

+
+## 🚫 No Cryptocurrency Affiliation
+
+Agentic Security is focused solely on AI security and has no affiliation with cryptocurrency projects, blockchain technologies, or related initiatives. Our mission is to advance the safety and reliability of AI systems—no tokens, no coins, just code.
+
 ## Contact us
@@ -1,3 +1,3 @@
-from .lib import AgenticSecurity
+from .lib import SecurityScanner

-__all__ = ["AgenticSecurity"]
+__all__ = ["SecurityScanner"]
@@ -5,7 +5,7 @@ import fire
 import uvicorn

 from agentic_security.app import app
-from agentic_security.lib import AgenticSecurity
+from agentic_security.lib import SecurityScanner
 from agentic_security.misc.banner import init_banner


@@ -32,14 +32,14 @@ class CLI:
        Run Agentic Security in CI mode.
        """
        sys.path.append(os.path.dirname("."))
-        AgenticSecurity().entrypoint()
+        SecurityScanner().entrypoint()

    def init(self, host: str = "0.0.0.0", port: int = 8718):
        """
        Generate the default CI configuration file.
        """
        sys.path.append(os.path.dirname("."))
-        AgenticSecurity().generate_default_settings(host, port)
+        SecurityScanner().generate_default_settings(host, port)

    i = init

@@ -48,7 +48,7 @@ class CLI:
        List all available security checks.
        """
        sys.path.append(os.path.dirname("."))
-        AgenticSecurity().list_checks()
+        SecurityScanner().list_checks()


 def main():
@@ -1,5 +1,4 @@
 import asyncio
-import logging
 import os
 from typing import Any

@@ -10,12 +9,11 @@ from pydantic import BaseModel, ConfigDict, Field

 # Assuming LLMSpec is defined elsewhere; placeholder import
 from agentic_security.http_spec import LLMSpec
+from agentic_security.logutils import logger

 LLM_SPECS = []  # Populate with LLM spec strings if needed

 # Configure logging
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger(__name__)


 # Define AgentSpecification model
@@ -1,5 +1,4 @@
 import asyncio
-import logging
 from typing import Any

 import httpx
@@ -8,13 +7,10 @@ from pydantic_ai import Agent, RunContext, Tool

 # Assuming LLMSpec is defined elsewhere; placeholder import
 from agentic_security.http_spec import LLMSpec
+from agentic_security.logutils import logger

 LLM_SPECS = []  # Populate this list with LLM spec strings if needed

-# Configure logging
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger(__name__)
-

 # Define AgentSpecification model
 class AgentSpecification(BaseModel):
@@ -1,7 +1,8 @@
 from functools import lru_cache

 import tomli
-from loguru import logger
+
+from agentic_security.logutils import logger

 SETTINGS_VERSION = 1

@@ -4,10 +4,14 @@ from asyncio import Event, Queue
 from fastapi import FastAPI
 from fastapi.responses import ORJSONResponse

+from agentic_security.http_spec import LLMSpec
+
 tools_inbox: Queue = Queue()
 stop_event: Event = Event()
 current_run: str = {"spec": "", "id": ""}
-_secrets = {}
+_secrets: dict[str, str] = {}
+
+current_run: dict[str, int | LLMSpec] = {"spec": "", "id": ""}


 def create_app() -> FastAPI:
@@ -26,29 +30,29 @@ def get_stop_event() -> Event:
    return stop_event


-def get_current_run() -> str:
+def get_current_run() -> dict[str, int | LLMSpec]:
    """Get the current run id."""
    return current_run


-def set_current_run(spec):
+def set_current_run(spec: LLMSpec) -> dict[str, int | LLMSpec]:
    """Set the current run id."""
    current_run["id"] = hash(id(spec))
    current_run["spec"] = spec
    return current_run


-def get_secrets():
+def get_secrets() -> dict[str, str]:
    return _secrets


-def set_secrets(secrets):
+def set_secrets(secrets: dict[str, str]) -> dict[str, str]:
    _secrets.update(secrets)
    expand_secrets(_secrets)
    return _secrets


-def expand_secrets(secrets):
+def expand_secrets(secrets: dict[str, str]) -> None:
    for key in secrets:
        val = secrets[key]
        if val.startswith("$"):
@@ -1,26 +1,5 @@
-from logging import config
+from agentic_security.logutils import set_log_level_to_info


 def setup_logging():
-    config.dictConfig(
-        {
-            "version": 1,
-            "disable_existing_loggers": True,
-            "handlers": {
-                "console": {
-                    "class": "logging.StreamHandler",
-                },
-            },
-            "root": {
-                "handlers": ["console"],
-                "level": "INFO",
-            },
-            "loggers": {
-                "uvicorn.access": {
-                    "level": "ERROR",  # Set higher log level to suppress info logs globally
-                    "handlers": ["console"],
-                    "propagate": False,
-                }
-            },
-        }
-    )
+    return set_log_level_to_info()
@@ -1,5 +1,6 @@
 import base64
 from enum import Enum
+from urllib.parse import urlparse

 import httpx
 from pydantic import BaseModel
@@ -159,6 +160,14 @@ def parse_http_spec(http_spec: str) -> LLMSpec:
    # Extract the method and URL from the first line
    method, url = lines[0].split(" ")[0:2]

+    # Check url validity
+    valid_url = urlparse(url)
+    # if missing the correct formatting ://, urlparse.netloc will be empty
+    if valid_url.scheme not in ("http", "https") or not valid_url.netloc:
+        raise InvalidHTTPSpecError(
+            f"Invalid URL: {url}. Ensure it starts with 'http://' or 'https://'"
+        )
+
    # Initialize headers and body
    headers = {}
    body = ""
@@ -4,12 +4,12 @@ from datetime import datetime

 import colorama
 import tqdm.asyncio
-from loguru import logger
 from rich.console import Console
 from rich.table import Table
 from tabulate import tabulate

 from agentic_security.config import SettingsMixin  # Importing the configuration mixin
+from agentic_security.logutils import logger
 from agentic_security.primitives import Scan
 from agentic_security.probe_data import REGISTRY
 from agentic_security.routes.scan import streaming_response_generator
@@ -23,7 +23,7 @@ YELLOW = colorama.Fore.YELLOW
 BLUE = colorama.Fore.BLUE


-class AgenticSecurity(SettingsMixin):
+class SecurityScanner(SettingsMixin):
    @classmethod
    async def async_scan(
        cls,
@@ -202,7 +202,7 @@ class AgenticSecurity(SettingsMixin):
        self.load_config(self.default_path)
        logger.info("Configuration loaded successfully.")
        print(self.config)
-        datasets = list(self.get_config_value("modules").values())
+        datasets = list(self.get_config_value("modules", {}).values())
        for d in datasets:
            d["selected"] = True
        self.scan(
@@ -0,0 +1,85 @@
+# import sys
+
+# from loguru import logger
+
+# # Define custom colors
+# BLUE = "#89CFF0"
+# BROWN = "#8B4513"  # Brown for DEBUG
+
+# # Define custom log level colors
+# logger.level("DEBUG", color=f"<fg {BROWN}>")
+# logger.level("INFO", color=f"<fg {BLUE}>")
+
+# # Define custom log format with aligned messages and colored levels
+# LOG_FORMAT = (
+#     "<level>{level:<8}</level> "  # Properly formatted and colored log level
+#     "<level>{message:<100}</level> "  # Left-aligned message for readability
+#     "<cyan>{file.name}</cyan>:<cyan>{line}</cyan>"  # File name and line number in cyan
+# )
+
+# # Remove default handlers and add a new one with custom formatting
+# logger.remove()
+# logger.add(sys.stdout, format=LOG_FORMAT, level="DEBUG", colorize=True)
+import logging
+import logging.config
+from os import getenv
+
+LOGGER_NAME = None
+
+LOGGING_CONFIG = {
+    "version": 1,
+    "disable_existing_loggers": False,
+    "formatters": {
+        "rich": {"format": "%(message)s", "datefmt": "[%X]"},
+    },
+    "handlers": {
+        "rich": {
+            "class": "rich.logging.RichHandler",
+            "level": "INFO",
+            "formatter": "rich",
+            "show_time": False,
+            "rich_tracebacks": False,
+            "show_path": lambda: True if getenv("API_RUNTIME") == "dev" else False,
+            "tracebacks_show_locals": False,
+        },
+    },
+    "loggers": {
+        "": {  # Root logger configuration
+            "level": "INFO",
+            "handlers": ["rich"],
+            "propagate": True,
+        },
+    },
+}
+
+
+def configure_logging():
+    # Apply the dictionary configuration
+    logging.config.dictConfig(LOGGING_CONFIG)
+
+    # Get and return the logger
+    logger = logging.getLogger(LOGGER_NAME)
+    return logger
+
+
+logger: logging.Logger = configure_logging()
+
+
+def set_log_level_to_debug():
+    logger = logging.getLogger(LOGGER_NAME)
+    logger.setLevel(logging.DEBUG)
+    # Update handler level as well
+    for handler in logger.handlers:
+        handler.setLevel(logging.DEBUG)
+
+
+def set_log_level_to_info():
+    logger = logging.getLogger(LOGGER_NAME)
+    logger.setLevel(logging.INFO)
+    # Update handler level as well
+    for handler in logger.handlers:
+        handler.setLevel(logging.INFO)
+
+
+# Set initial log level
+set_log_level_to_info()
@@ -1,7 +1,8 @@
 from fastapi import Request
-from loguru import logger
 from starlette.middleware.base import BaseHTTPMiddleware

+from agentic_security.logutils import logger
+

 class LogNon200ResponsesMiddleware(BaseHTTPMiddleware):
    async def dispatch(self, request: Request, call_next):
@@ -8,14 +8,14 @@ except ImportError:


 def generate_banner(
-    title="Agentic Security",
-    font="slant",
-    version="v2.1.0",
-    tagline="Proactive Threat Detection & Automated Security Protocols",
-    author="Developed by: [Security Team]",
-    website="Website: https://github.com/msoedov/agentic_security",
-    warning="",
-):
+    title: str = "Agentic Security",
+    font: str = "slant",
+    version: str = "v2.1.0",
+    tagline: str = "Proactive Threat Detection & Automated Security Protocols",
+    author: str = "Developed by: [Security Team]",
+    website: str = "Website: https://github.com/msoedov/agentic_security",
+    warning: str | None = "",  # Using Optional for warning since it might be None
+) -> str:
    """Generate a visually enhanced banner with dynamic width and borders."""
    # Define the text elements

@@ -80,8 +80,13 @@ def generate_banner(


 def init_banner():
+    return
    ver = version("agentic_security")
-    print(generate_banner(version=ver))
+    try:
+        print(generate_banner(version=ver))
+    except Exception:
+        # UnicodeEncodeError with codec on some systems
+        pass


 if __name__ == "__main__":
@@ -6,11 +6,11 @@ from json import JSONDecodeError

 import httpx
 import pandas as pd
-from loguru import logger
 from skopt import Optimizer
 from skopt.space import Real

 from agentic_security.http_spec import Modality
+from agentic_security.logutils import logger
 from agentic_security.primitives import Scan, ScanResult
 from agentic_security.probe_actor.cost_module import calculate_cost
 from agentic_security.probe_actor.refusal import refusal_heuristic
@@ -107,6 +107,17 @@ async def process_prompt_batch(
    return total_tokens, failures


+async def with_error_handling(agen):
+    try:
+        async for t in agen:
+            yield t
+    except Exception as e:
+        logger.exception("Scan failed")
+        yield ScanResult.status_msg(f"Scan failed: {str(e)}")
+    finally:
+        yield ScanResult.status_msg("Scan completed.")
+
+
 async def perform_single_shot_scan(
    request_factory,
    max_budget: int,
@@ -120,125 +131,117 @@ async def perform_single_shot_scan(
    max_budget = max_budget * BUDGET_MULTIPLIER
    selected_datasets = [m for m in datasets if m["selected"]]
    request_factory = multi_modality_spec(request_factory)
-    try:
-        yield ScanResult.status_msg("Loading datasets...")
-        prompt_modules = prepare_prompts(
-            dataset_names=[m["dataset_name"] for m in selected_datasets],
-            budget=max_budget,
-            tools_inbox=tools_inbox,
-            options=[m.get("opts", {}) for m in selected_datasets],
-        )
-        yield ScanResult.status_msg("Datasets loaded. Starting scan...")
+    yield ScanResult.status_msg("Loading datasets...")
+    prompt_modules = prepare_prompts(
+        dataset_names=[m["dataset_name"] for m in selected_datasets],
+        budget=max_budget,
+        tools_inbox=tools_inbox,
+        options=[m.get("opts", {}) for m in selected_datasets],
+    )
+    yield ScanResult.status_msg("Datasets loaded. Starting scan...")

-        errors = []
-        refusals = []
-        outputs = []
-        total_prompts = sum(len(m.prompts) for m in prompt_modules if not m.lazy)
-        processed_prompts = 0
+    errors = []
+    refusals = []
+    outputs = []
+    total_prompts = sum(len(m.prompts) for m in prompt_modules if not m.lazy)
+    processed_prompts = 0

-        optimizer = (
-            Optimizer([Real(0, 1)], base_estimator="GP", n_initial_points=25)
-            if optimize
-            else None
-        )
-        failure_rates = []
+    optimizer = (
+        Optimizer([Real(0, 1)], base_estimator="GP", n_initial_points=25)
+        if optimize
+        else None
+    )
+    failure_rates = []

-        total_tokens = 0
+    total_tokens = 0
+    tokens = 0
+    should_stop = False
+    for module in prompt_modules:
+        if should_stop:
+            break
        tokens = 0
-        should_stop = False
-        for module in prompt_modules:
-            if should_stop:
-                break
-            tokens = 0
-            module_failures = 0
-            module_size = 0 if module.lazy else len(module.prompts)
-            logger.info(f"Scanning {module.dataset_name} {module_size}")
-            module_prompts = 0  # Reset for each module
+        module_failures = 0
+        module_size = 0 if module.lazy else len(module.prompts)
+        logger.info(f"Scanning {module.dataset_name} {module_size}")
+        module_prompts = 0  # Reset for each module

-            async for prompt in generate_prompts(module.prompts):
-                if stop_event and stop_event.is_set():
-                    stop_event.clear()
-                    logger.info("Scan stopped by user.")
-                    yield ScanResult.status_msg("Scan stopped by user.")
-                    return
+        async for prompt in generate_prompts(module.prompts):
+            if stop_event and stop_event.is_set():
+                stop_event.clear()
+                logger.info("Scan stopped by user.")
+                yield ScanResult.status_msg("Scan stopped by user.")
+                return

-                processed_prompts += 1
-                module_prompts += 1  # Fixed increment syntax
-                # Calculate progress based on total processed prompts
-                progress = (
-                    100 * processed_prompts / total_prompts if total_prompts else 0
-                )
+            processed_prompts += 1
+            module_prompts += 1  # Fixed increment syntax
+            # Calculate progress based on total processed prompts
+            progress = 100 * processed_prompts / total_prompts if total_prompts else 0
+            progress = progress % 100

-                total_tokens -= tokens
-                start = time.time()
-                tokens, failed = await process_prompt(
-                    request_factory,
-                    prompt,
-                    tokens,
-                    module.dataset_name,
-                    refusals,
-                    errors,
-                    outputs,
-                )
-                end = time.time()
-                total_tokens += tokens
+            total_tokens -= tokens
+            start = time.time()
+            tokens, failed = await process_prompt(
+                request_factory,
+                prompt,
+                tokens,
+                module.dataset_name,
+                refusals,
+                errors,
+                outputs,
+            )
+            end = time.time()
+            total_tokens += tokens

-                if failed:
-                    module_failures += 1
-                failure_rate = module_failures / max(module_prompts, 1)
-                failure_rates.append(failure_rate)
-                cost = calculate_cost(tokens)
+            if failed:
+                module_failures += 1
+            failure_rate = module_failures / max(module_prompts, 1)
+            failure_rates.append(failure_rate)
+            cost = calculate_cost(tokens)

-                last_output = outputs[-1] if outputs else None
-                if last_output and last_output[1] == prompt:
-                    response_text = last_output[2]
-                else:
-                    response_text = ""
+            last_output = outputs[-1] if outputs else None
+            if last_output and last_output[1] == prompt:
+                response_text = last_output[2]
+            else:
+                response_text = ""

-                yield ScanResult(
-                    module=module.dataset_name,
-                    tokens=round(tokens / 1000, 1),
-                    cost=cost,
-                    progress=round(progress, 2),
-                    failureRate=round(failure_rate * 100, 2),
-                    prompt=prompt[:MAX_PROMPT_LENGTH],
-                    latency=end - start,
-                    model=response_text,
-                ).model_dump_json()
+            yield ScanResult(
+                module=module.dataset_name,
+                tokens=round(tokens / 1000, 1),
+                cost=cost,
+                progress=round(progress, 2),
+                failureRate=round(failure_rate * 100, 2),
+                prompt=prompt[:MAX_PROMPT_LENGTH],
+                latency=end - start,
+                model=response_text,
+            ).model_dump_json()

-                if optimize and len(failure_rates) >= 5:
-                    next_point = optimizer.ask()
-                    optimizer.tell(next_point, -failure_rate)
-                    best_failure_rate = -optimizer.get_result().fun
-                    if best_failure_rate > 0.5:
-                        yield ScanResult.status_msg(
-                            f"High failure rate detected ({best_failure_rate:.2%}). Stopping this module..."
-                        )
-                        should_stop = True
-                        break
-                if total_tokens > max_budget:
-                    logger.info(
-                        f"Scan ran out of budget and stopped. {total_tokens=} {max_budget=}"
-                    )
+            if optimize and len(failure_rates) >= 5:
+                next_point = optimizer.ask()
+                optimizer.tell(next_point, -failure_rate)
+                best_failure_rate = -optimizer.get_result().fun
+                if best_failure_rate > 0.5:
                    yield ScanResult.status_msg(
-                        f"Scan ran out of budget and stopped. {total_tokens=} {max_budget=}"
+                        f"High failure rate detected ({best_failure_rate:.2%}). Stopping this module..."
                    )
                    should_stop = True
                    break
+            if total_tokens > max_budget:
+                logger.info(
+                    f"Scan ran out of budget and stopped. {total_tokens=} {max_budget=}"
+                )
+                yield ScanResult.status_msg(
+                    f"Scan ran out of budget and stopped. {total_tokens=} {max_budget=}"
+                )
+                should_stop = True
+                break

-        yield ScanResult.status_msg("Scan completed.")
+    yield ScanResult.status_msg("Scan completed.")

-        failure_data = errors + refusals
-        df = pd.DataFrame(
-            failure_data, columns=["module", "prompt", "status_code", "content"]
-        )
-        df.to_csv("failures.csv", index=False)
-
-    except Exception as e:
-        logger.exception("Scan failed")
-        yield ScanResult.status_msg(f"Scan failed: {str(e)}")
-    finally:
-        yield ScanResult.status_msg("Scan completed.")
+    failure_data = errors + refusals
+    df = pd.DataFrame(
+        failure_data, columns=["module", "prompt", "status_code", "content"]
+    )
+    df.to_csv("failures.csv", index=False)


 async def perform_many_shot_scan(
@@ -255,114 +258,107 @@ async def perform_many_shot_scan(
 ) -> AsyncGenerator[str, None]:
    """Perform a multi-step security scan with probe injection."""
    request_factory = multi_modality_spec(request_factory)
-    try:
-        # Load main and probe datasets
-        yield ScanResult.status_msg("Loading datasets...")
-        prompt_modules = prepare_prompts(
-            dataset_names=[m["dataset_name"] for m in datasets if m["selected"]],
-            budget=max_budget,
-            tools_inbox=tools_inbox,
-        )
-        yield ScanResult.status_msg("Loading datasets for MSJ...")
-        msj_modules = msj_data.prepare_prompts(probe_datasets)
-        yield ScanResult.status_msg("Datasets loaded. Starting scan...")
+    # Load main and probe datasets
+    yield ScanResult.status_msg("Loading datasets...")
+    prompt_modules = prepare_prompts(
+        dataset_names=[m["dataset_name"] for m in datasets if m["selected"]],
+        budget=max_budget,
+        tools_inbox=tools_inbox,
+    )
+    yield ScanResult.status_msg("Loading datasets for MSJ...")
+    msj_modules = msj_data.prepare_prompts(probe_datasets)
+    yield ScanResult.status_msg("Datasets loaded. Starting scan...")

-        errors = []
-        refusals = []
-        outputs = []
-        total_prompts = sum(len(m.prompts) for m in prompt_modules if not m.lazy)
-        processed_prompts = 0
+    errors = []
+    refusals = []
+    outputs = []
+    total_prompts = sum(len(m.prompts) for m in prompt_modules if not m.lazy)
+    processed_prompts = 0

-        optimizer = (
-            Optimizer([Real(0, 1)], base_estimator="GP", n_initial_points=25)
-            if optimize
-            else None
-        )
-        failure_rates = []
+    optimizer = (
+        Optimizer([Real(0, 1)], base_estimator="GP", n_initial_points=25)
+        if optimize
+        else None
+    )
+    failure_rates = []

-        for module in prompt_modules:
-            module_failures = 0
-            module_size = 0 if module.lazy else len(module.prompts)
-            logger.info(f"Scanning {module.dataset_name} {module_size}")
+    for module in prompt_modules:
+        module_failures = 0
+        module_size = 0 if module.lazy else len(module.prompts)
+        logger.info(f"Scanning {module.dataset_name} {module_size}")

-            async for prompt in generate_prompts(module.prompts):
-                if stop_event and stop_event.is_set():
-                    stop_event.clear()
-                    logger.info("Scan stopped by user.")
-                    yield ScanResult.status_msg("Scan stopped by user.")
-                    return
-                tokens = 0
-                processed_prompts += 1
-                progress = (
-                    100 * processed_prompts / total_prompts if total_prompts else 0
+        async for prompt in generate_prompts(module.prompts):
+            if stop_event and stop_event.is_set():
+                stop_event.clear()
+                logger.info("Scan stopped by user.")
+                yield ScanResult.status_msg("Scan stopped by user.")
+                return
+            tokens = 0
+            processed_prompts += 1
+            progress = 100 * processed_prompts / total_prompts if total_prompts else 0
+            progress = progress % 100
+
+            full_prompt = ""
+            msj_module = random.choice(msj_modules)
+
+            prompt_tokens = len(full_prompt.split())
+            tokens += prompt_tokens
+
+            injected = False
+            for _ in range(20):
+                if injected:
+                    break
+
+                m_prompt = random.choice(msj_module.prompts)
+                full_prompt += "\n" + m_prompt
+                if tokens > max_ctx_length:
+                    full_prompt = "\n" + prompt
+                    injected = True
+
+                tokens, failed = await process_prompt(
+                    request_factory,
+                    full_prompt,
+                    tokens,
+                    module.dataset_name,
+                    refusals,
+                    errors,
+                    outputs,
                )
+                if failed:
+                    module_failures += 1
+                    break
+                if injected:
+                    break

-                full_prompt = ""
-                msj_module = random.choice(msj_modules)
+            failure_rate = module_failures / max(processed_prompts, 1)
+            failure_rates.append(failure_rate)
+            cost = calculate_cost(tokens)

-                prompt_tokens = len(full_prompt.split())
-                tokens += prompt_tokens
+            yield ScanResult(
+                module=module.dataset_name,
+                tokens=round(tokens / 1000, 1),
+                cost=cost,
+                progress=round(progress, 2),
+                failureRate=round(failure_rate * 100, 2),
+                prompt=prompt[:MAX_PROMPT_LENGTH],
+            ).model_dump_json()

-                injected = False
-                for _ in range(20):
-                    if injected:
-                        break
-
-                    m_prompt = random.choice(msj_module.prompts)
-                    full_prompt += "\n" + m_prompt
-                    if tokens > max_ctx_length:
-                        full_prompt = "\n" + prompt
-                        injected = True
-
-                    tokens, failed = await process_prompt(
-                        request_factory,
-                        full_prompt,
-                        tokens,
-                        module.dataset_name,
-                        refusals,
-                        errors,
-                        outputs,
+            if optimize and len(failure_rates) >= 5:
+                next_point = optimizer.ask()
+                optimizer.tell(next_point, -failure_rate)
+                best_failure_rate = -optimizer.get_result().fun
+                if best_failure_rate > 0.5:
+                    yield ScanResult.status_msg(
+                        f"High failure rate detected ({best_failure_rate:.2%}). Stopping this module..."
                    )
-                    if failed:
-                        module_failures += 1
-                        break
-                    if injected:
-                        break
+                    break

-                failure_rate = module_failures / max(processed_prompts, 1)
-                failure_rates.append(failure_rate)
-                cost = calculate_cost(tokens)
+    yield ScanResult.status_msg("Scan completed.")

-                yield ScanResult(
-                    module=module.dataset_name,
-                    tokens=round(tokens / 1000, 1),
-                    cost=cost,
-                    progress=round(progress, 2),
-                    failureRate=round(failure_rate * 100, 2),
-                    prompt=prompt[:MAX_PROMPT_LENGTH],
-                ).model_dump_json()
-
-                if optimize and len(failure_rates) >= 5:
-                    next_point = optimizer.ask()
-                    optimizer.tell(next_point, -failure_rate)
-                    best_failure_rate = -optimizer.get_result().fun
-                    if best_failure_rate > 0.5:
-                        yield ScanResult.status_msg(
-                            f"High failure rate detected ({best_failure_rate:.2%}). Stopping this module..."
-                        )
-                        break
-
-        yield ScanResult.status_msg("Scan completed.")
-
-        df = pd.DataFrame(
-            errors + refusals, columns=["module", "prompt", "status_code", "content"]
-        )
-        df.to_csv("failures.csv", index=False)
-
-    except Exception as e:
-        logger.exception("Scan failed")
-        yield ScanResult.status_msg(f"Scan failed: {str(e)}")
-        raise e
+    df = pd.DataFrame(
+        errors + refusals, columns=["module", "prompt", "status_code", "content"]
+    )
+    df.to_csv("failures.csv", index=False)


 def scan_router(
@@ -372,23 +368,27 @@ def scan_router(
    stop_event: asyncio.Event = None,
 ):
    if scan_parameters.enableMultiStepAttack:
-        return perform_many_shot_scan(
-            request_factory=request_factory,
-            max_budget=scan_parameters.maxBudget,
-            datasets=scan_parameters.datasets,
-            probe_datasets=scan_parameters.probe_datasets,
-            tools_inbox=tools_inbox,
-            optimize=scan_parameters.optimize,
-            stop_event=stop_event,
-            secrets=scan_parameters.secrets,
+        return with_error_handling(
+            perform_many_shot_scan(
+                request_factory=request_factory,
+                max_budget=scan_parameters.maxBudget,
+                datasets=scan_parameters.datasets,
+                probe_datasets=scan_parameters.probe_datasets,
+                tools_inbox=tools_inbox,
+                optimize=scan_parameters.optimize,
+                stop_event=stop_event,
+                secrets=scan_parameters.secrets,
+            )
        )
    else:
-        return perform_single_shot_scan(
-            request_factory=request_factory,
-            max_budget=scan_parameters.maxBudget,
-            datasets=scan_parameters.datasets,
-            tools_inbox=tools_inbox,
-            optimize=scan_parameters.optimize,
-            stop_event=stop_event,
-            secrets=scan_parameters.secrets,
+        return with_error_handling(
+            perform_single_shot_scan(
+                request_factory=request_factory,
+                max_budget=scan_parameters.maxBudget,
+                datasets=scan_parameters.datasets,
+                tools_inbox=tools_inbox,
+                optimize=scan_parameters.optimize,
+                stop_event=stop_event,
+                secrets=scan_parameters.secrets,
+            )
        )
@@ -1,5 +1,4 @@
 import asyncio
-import logging
 from typing import Any

 import httpx
@@ -7,13 +6,10 @@ from pydantic import BaseModel, Field
 from pydantic_ai import Agent, RunContext

 from agentic_security.http_spec import LLMSpec
+from agentic_security.logutils import logger

 LLM_SPECS = []

-# Configure logging
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger(__name__)
-

 class AgentSpecification(BaseModel):
    name: str | None = Field(None, description="Name of the LLM/agent")
@@ -1,6 +1,6 @@
 from abc import ABC, abstractmethod

-from agentic_security.refusal_classifier import RefusalClassifier
+from agentic_security.refusal_classifier.model import RefusalClassifier

 classifier = RefusalClassifier()
 classifier.load_model()
@@ -1,4 +1,5 @@
 import base64
+import logging
 import os
 import platform
 import subprocess
@@ -7,6 +8,16 @@ import uuid
 import httpx
 from cache_to_disk import cache_to_disk

+# Configure logging
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+
+class AudioGenerationError(Exception):
+    """Custom exception for errors during audio generation."""
+
+    pass
+

 def encode(content: bytes) -> str:
    encoded_content = base64.b64encode(content).decode("utf-8")
@@ -41,12 +52,25 @@ def generate_audio_mac_wav(prompt: str) -> bytes:
        # Read the WAV file into memory
        with open(temp_wav_path, "rb") as f:
            audio_bytes = f.read()
+
+    except subprocess.CalledProcessError as e:
+        logger.error(f"Subprocess error: {e}")
+        raise AudioGenerationError("Failed to generate or convert audio.") from e
+    except FileNotFoundError as e:
+        logger.error(f"File not found: {e}")
+        raise AudioGenerationError("Required file not found.") from e
+    except Exception as e:
+        logger.exception("Unexpected error occurred.")
+        raise AudioGenerationError(
+            "An unexpected error occurred during audio generation."
+        ) from e
    finally:
-        # Clean up the temporary files
-        if os.path.exists(temp_aiff_path):
-            os.remove(temp_aiff_path)
-        if os.path.exists(temp_wav_path):
-            os.remove(temp_wav_path)
+        for path in (temp_aiff_path, temp_wav_path):
+            try:
+                if os.path.exists(path):
+                    os.remove(path)
+            except Exception as e:
+                logger.warning(f"Failed to delete temporary file {path}: {e}")

    # Return the audio bytes
    return audio_bytes
@@ -1,14 +1,15 @@
 import io
 import os
 import random
-from functools import lru_cache
+from collections.abc import Callable, Iterator
+from functools import partial

 import httpx
 import pandas as pd
 from cache_to_disk import cache_to_disk
 from datasets import load_dataset
-from loguru import logger

+from agentic_security.logutils import logger
 from agentic_security.probe_data import stenography_fn
 from agentic_security.probe_data.models import ProbeDataset
 from agentic_security.probe_data.modules import (
@@ -19,281 +20,200 @@ from agentic_security.probe_data.modules import (
    rl_model,
 )

+# Type aliases for clarity
+FilterFn = Callable[[pd.Series], bool]
+ColumnMappings = dict[str, str]
+DatasetLoader = Callable[[], ProbeDataset]

-@cache_to_disk()
-def load_dataset_general(
-    dataset_name,
-    dataset_split="train",
-    column_mappings=None,
-    filter_fn=None,
-    custom_url=None,
-    additional_metadata=None,
-):
-    """
-    Generalized function to load datasets with flexible configurations.

-    :param dataset_name: Name of the dataset or URL for custom CSVs
-    :param dataset_split: Split to load from the dataset (e.g., "train")
-    :param column_mappings: Dictionary mapping dataset columns to expected keys, e.g., {'prompt': 'query'}
-    :param filter_fn: A filtering function that takes a row and returns True/False
-    :param custom_url: URL for custom CSV datasets
-    :param additional_metadata: Additional metadata to include in the ProbeDataset
-    :return: A ProbeDataset object with the processed data
-    """
-    if custom_url:
-        logger.info(f"Loading custom CSV dataset from {custom_url}")
-        r = httpx.get(custom_url)
-        content = r.content
-        df = pd.read_csv(io.StringIO(content.decode("utf-8")))
-    else:
-        logger.info(f"Loading dataset {dataset_name} from Hugging Face datasets")
-        dataset = load_dataset(dataset_name)
-        df = pd.DataFrame(dataset[dataset_split])
+# Core data loading utilities
+def fetch_csv_content(url: str) -> str:
+    """Fetch CSV content from a URL."""
+    response = httpx.get(url)
+    return response.content.decode("utf-8")

-    # Apply column mappings if provided
-    if column_mappings:
-        df.rename(columns=column_mappings, inplace=True)

-    # Filter rows if filter_fn is provided
+def load_df_from_source(source: str, is_url: bool = False) -> pd.DataFrame:
+    """Load DataFrame from either URL or Hugging Face dataset."""
+    if is_url:
+        logger.info(f"Loading custom CSV from {source}")
+        return pd.read_csv(io.StringIO(fetch_csv_content(source)))
+    logger.info(f"Loading dataset {source} from Hugging Face")
+    return pd.DataFrame(load_dataset(source)["train"])
+
+
+def transform_df(
+    df: pd.DataFrame,
+    mappings: ColumnMappings | None = None,
+    filter_fn: FilterFn | None = None,
+) -> pd.DataFrame:
+    """Apply transformations to DataFrame."""
+    result = df.copy()
+    if mappings:
+        result.rename(columns=mappings, inplace=True)
    if filter_fn:
-        df = df[df.apply(filter_fn, axis=1)]
-
-    # Extract prompts
-    prompts = df[column_mappings.get("prompt", "prompt")].tolist()
-
-    return ProbeDataset(
-        dataset_name=dataset_name,
-        metadata=additional_metadata or {},
-        prompts=prompts,
-        tokens=count_words_in_list(prompts),
-        approx_cost=0.0,
-    )
+        result = result[result.apply(filter_fn, axis=1)]
+    return result


-def count_words_in_list(str_list):
-    """Calculate the total number of words in a given list of strings.
-
-    :param str_list: List of strings
-    :return: Total number of words across all strings in the list
-    """
-    total_words = sum(len(str(s).split()) for s in str_list)
-    return total_words
-
-
-@cache_to_disk()
-def load_dataset_v1():
-    return load_dataset_general(
-        dataset_name="ShawnMenz/DAN_jailbreak",
-        column_mappings={"prompt": "prompt", "jailbreak": "jailbreak"},
-        filter_fn=lambda row: row["jailbreak"],
-    )
-
-
-@cache_to_disk()
-def load_dataset_v2():
-    return load_dataset_general(
-        dataset_name="deepset/prompt-injections",
-        column_mappings={"prompt": "text", "jailbreak": "label"},
-        filter_fn=lambda row: row["label"],
-    )
-
-
-@cache_to_disk()
-def load_dataset_v4():
-    return load_dataset_general(
-        dataset_name="notrichardren/refuse-to-answer-prompts",
-        column_mappings={"prompt": "claim", "jailbreak": "label"},
-        filter_fn=lambda row: row["label"],
-    )
-
-
-@cache_to_disk()
-def load_dataset_v3():
-    return load_dataset_general(
-        dataset_name="rubend18/ChatGPT-Jailbreak-Prompts",
-        column_mappings={"prompt": "Prompt"},
-        filter_fn=lambda row: row["label"],
-    )
-
-
-@cache_to_disk()
-def load_dataset_v6():
-    dataset = load_dataset("markush1/LLM-Jailbreak-Classifier")
-    bad_prompts = [
-        p
-        for p, c in zip(dataset["train"]["prompt"], dataset["train"]["classification"])
-        if c == "jailbreak"
-    ]
-    return ProbeDataset(
-        dataset_name="markush1/LLM-Jailbreak-Classifier",
-        metadata={},
-        prompts=bad_prompts,
-        tokens=count_words_in_list(bad_prompts),
-        approx_cost=0.0,
-    )
-
-
-@cache_to_disk()
-def load_dataset_v7():
-    splits = {
-        "mini_JailBreakV_28K": "JailBreakV_28K/mini_JailBreakV_28K.csv",
-        "JailBreakV_28K": "JailBreakV_28K/JailBreakV_28K.csv",
-    }
-    df = pd.read_csv(
-        "hf://datasets/JailbreakV-28K/JailBreakV-28k/" + splits["JailBreakV_28K"]
-    )
-    bad_prompts = df["jailbreak_query"].tolist()
-    print(df.shape)
-    return ProbeDataset(
-        dataset_name="JailbreakV-28K/JailBreakV-28k",
-        metadata={},
-        prompts=bad_prompts,
-        tokens=count_words_in_list(bad_prompts),
-        approx_cost=0.0,
-    )
-
-
-@cache_to_disk()
-def load_dataset_v8():
-    df = pd.read_csv(
-        "hf://datasets/ShawnMenz/jailbreak_sft_rm_ds/jailbreak_sft_rm_ds.csv",
-        names=["jailbreak", "prompt"],
-    )
-    filtered = df[df["jailbreak"] == "jailbreak"]["prompt"].tolist()
-    return ProbeDataset(
-        dataset_name="JailbreakV-28K/JailBreakV-28k",
-        metadata={},
-        prompts=filtered,
-        tokens=count_words_in_list(filtered),
-        approx_cost=0.0,
-    )
-
-
-@cache_to_disk()
-def load_dataset_v5():
-    ds = []
-    for c in [
-        "AdvBench",
-        "ForbiddenQuestion",
-        "MJP",
-        "MaliciousInstruct",
-        "QuestionList",
-    ]:
-        dataset = load_dataset("Lemhf14/EasyJailbreak_Datasets", c)
-        bad_prompts = dataset["train"]["query"]
-        ds.extend(bad_prompts)
-
-    return ProbeDataset(
-        dataset_name="Lemhf14/EasyJailbreak_Datasets",
-        metadata={},
-        prompts=ds,
-        tokens=count_words_in_list(ds),
-        approx_cost=0.0,
-    )
-
-
-@cache_to_disk()
-def load_generic_csv(url, name, column="prompt", predicator=None):
-    r = httpx.get(url)
-    content = r.content
-    df = pd.read_csv(io.StringIO(content.decode("utf-8")))
-    logger.info(f"Loaded {len(df)} prompts from {url}")
-    filtered_prompts = df[df.apply(predicator, axis=1)][column].tolist()
+def create_probe_dataset(
+    name: str, prompts: list[str], metadata: dict = None
+) -> ProbeDataset:
+    """Create a ProbeDataset from prompts."""
+    metadata = metadata or {}
    return ProbeDataset(
        dataset_name=name,
-        metadata={},
-        prompts=filtered_prompts,
-        tokens=count_words_in_list(filtered_prompts),
+        metadata=metadata,
+        prompts=prompts,
+        tokens=sum(len(str(s).split()) for s in prompts),
        approx_cost=0.0,
    )


-def prepare_prompts(dataset_names, budget, tools_inbox=None, options=[]):
-    # ## Datasets used and cleaned:
-    # markush1/LLM-Jailbreak-Classifier
-    # 1. Open-Orca/OpenOrca
-    # 2. ShawnMenz/DAN_jailbreak
-    # 3. EddyLuo/JailBreakV_28K
-    # 4. https://raw.githubusercontent.com/verazuo/jailbreak_llms/main/data/jailbreak_prompts.csv
-
-    dataset_map = {
-        "ShawnMenz/DAN_jailbreak": load_dataset_v1,
-        "deepset/prompt-injections": load_dataset_v2,
-        "notrichardren/refuse-to-answer-prompts": load_dataset_v4,
-        "rubend18/ChatGPT-Jailbreak-Prompts": load_dataset_v3,
-        "Lemhf14/EasyJailbreak_Datasets": load_dataset_v5,
-        "markush1/LLM-Jailbreak-Classifier": load_dataset_v6,
-        "JailbreakV-28K/JailBreakV-28k": load_dataset_v7,
-        "ShawnMenz/jailbreak_sft_rm_ds": load_dataset_v8,
-        "verazuo/jailbreak_llms/2023_05_07": lambda: load_generic_csv(
-            url="https://raw.githubusercontent.com/verazuo/jailbreak_llms/main/data/prompts/jailbreak_prompts_2023_05_07.csv",
-            name="verazuo/jailbreak_llms/2023_05_07",
-            column="prompt",
-            predicator=lambda x: bool(x["jailbreak"]),
-        ),
-        "verazuo/jailbreak_llms/2023_12_25.csv": lambda: load_generic_csv(
-            url="https://raw.githubusercontent.com/verazuo/jailbreak_llms/main/data/prompts/jailbreak_prompts_2023_12_25.csv.csv",
-            name="verazuo/jailbreak_llms/2023_12_25.csv",
-            column="prompt",
-            predicator=lambda x: bool(x["jailbreak"]),
-        ),
-        "Custom CSV": load_local_csv,
-    }
-    dataset_map.update(dataset_map_generics)
-
-    group = []
-    for dataset_name in dataset_names:
-        if dataset_name in dataset_map:
-            logger.info(f"Loading {dataset_name}")
-            try:
-                group.append(dataset_map[dataset_name]())
-            except Exception as e:
-                logger.error(f"Error loading {dataset_name}: {e}")
-
-    dynamic_datasets = {
-        "AgenticBackend": lambda opts: dataset_from_iterator(
-            "AgenticBackend",
-            fine_tuned.Module(group, tools_inbox=tools_inbox, opts=opts).apply(),
-            lazy=True,
-        ),
-        "Steganography": lambda opts: Stenography(group),
-        "llm-adaptive-attacks": lambda opts: dataset_from_iterator(
-            "llm-adaptive-attacks",
-            adaptive_attacks.Module(group, tools_inbox=tools_inbox, opts=opts).apply(),
-        ),
-        "Garak": lambda opts: dataset_from_iterator(
-            "Garak",
-            garak_tool.Module(group, tools_inbox=tools_inbox, opts=opts).apply(),
-            lazy=True,
-        ),
-        "Reinforcement Learning Optimization": lambda opts: dataset_from_iterator(
-            "Reinforcement Learning Optimization",
-            rl_model.Module(group, tools_inbox=tools_inbox, opts=opts).apply(),
-            lazy=True,
-        ),
-        "InspectAI": lambda opts: dataset_from_iterator(
-            "InspectAI",
-            inspect_ai_tool.Module(group, tools_inbox=tools_inbox).apply(),
-            lazy=True,
-        ),
-        "GPT fuzzer": lambda opts: [],
-    }
-
-    dynamic_groups = []
-    options = options or [{} for _ in dataset_names]
-    for dataset_name, opts in zip(dataset_names, options):
-        if dataset_name in dynamic_datasets:
-            logger.info(f"Loading {dataset_name}")
-
-            ds = dynamic_datasets[dataset_name](opts)
-
-            for g in ds:
-                dynamic_groups.append(g)
-    return group + dynamic_groups
+# Generalized dataset loader
+@cache_to_disk()
+def load_dataset_generic(
+    name: str,
+    mappings: ColumnMappings | None = None,
+    filter_fn: FilterFn | None = None,
+    url: str | None = None,
+    metadata: dict | None = None,
+) -> ProbeDataset:
+    """Load and process a dataset with flexible configuration."""
+    df = load_df_from_source(url or name, is_url=bool(url))
+    transformed_df = transform_df(df, mappings, filter_fn)
+    prompt_col = mappings.get("prompt", "prompt") if mappings else "prompt"
+    prompts = transformed_df[prompt_col].tolist()
+    return create_probe_dataset(name, prompts, metadata)


-class Stenography:
-    fn_library = {
+# Dataset-specific configurations
+DATASET_CONFIGS = {
+    "ShawnMenz/DAN_jailbreak": {
+        "mappings": {"prompt": "prompt"},
+        "filter_fn": lambda row: row["jailbreak"],
+    },
+    "deepset/prompt-injections": {
+        "mappings": {"prompt": "text"},
+        "filter_fn": lambda row: row["label"],
+    },
+    "notrichardren/refuse-to-answer-prompts": {
+        "mappings": {"prompt": "claim"},
+        "filter_fn": lambda row: row["label"],
+    },
+    "rubend18/ChatGPT-Jailbreak-Prompts": {
+        "mappings": {"prompt": "Prompt"},
+        "filter_fn": lambda row: row["label"],
+    },
+    "markush1/LLM-Jailbreak-Classifier": {
+        "mappings": {"prompt": "prompt"},
+        "filter_fn": lambda row: row["classification"] == "jailbreak",
+    },
+    "ShawnMenz/jailbreak_sft_rm_ds": {
+        "url": "hf://datasets/ShawnMenz/jailbreak_sft_rm_ds/jailbreak_sft_rm_ds.csv",
+        "mappings": {"prompt": "prompt"},
+        "filter_fn": lambda row: row["jailbreak"] == "jailbreak",
+    },
+    "verazuo/jailbreak_llms/2023_05_07": {
+        "url": "https://raw.githubusercontent.com/verazuo/jailbreak_llms/main/data/prompts/jailbreak_prompts_2023_05_07.csv",
+        "mappings": {"prompt": "prompt"},
+        "filter_fn": lambda row: bool(row["jailbreak"]),
+    },
+    "verazuo/jailbreak_llms/2023_12_25": {
+        "url": "https://raw.githubusercontent.com/verazuo/jailbreak_llms/main/data/prompts/jailbreak_prompts_2023_12_25.csv",
+        "mappings": {"prompt": "prompt"},
+        "filter_fn": lambda row: bool(row["jailbreak"]),
+    },
+}
+
+# Additional generic dataset configurations
+DATASET_CONFIGS_GENERICS = {
+    "simonycl/aya-23-8B_advbench_jailbreak": {"mappings": {"prompt": "prompt"}},
+    "acmc/jailbreaks_dataset_with_perplexity_bigcode_starcoder2-3b_bigcode_starcoder2-7b": {},
+    "karanxa/dolphin-jailbreak-finetuning-dataset": {"mappings": {"prompt": "text"}},
+    "karanxa/llama-2-jailbreak-dataset": {"mappings": {"prompt": "text"}},
+    "karanxa/llama2-uncensored-jailbreak-dataset-finetuning": {
+        "mappings": {"prompt": "text"}
+    },
+    "liuyanchen1015/Llama-3.2-1B_jailbreak_responses": {
+        "mappings": {"prompt": "jailbreak_prompt_text"}
+    },
+    "liuyanchen1015/Llama-3.2-1B-Instruct_jailbreak_responses": {
+        "mappings": {"prompt": "jailbreak_prompt_text"}
+    },
+    "liuyanchen1015/Llama-3.2-1B-Instruct_jailbreak_responses_with_judgment": {
+        "mappings": {"prompt": "jailbreak_prompt_text"}
+    },
+    "jackhhao/jailbreak-classification": {"mappings": {"prompt": "prompt"}},
+    "walledai/JailbreakBench": {"mappings": {"prompt": "prompt"}},
+    "walledai/JailbreakHub": {"mappings": {"prompt": "prompt"}},
+    "Granther/evil-jailbreak": {"mappings": {"prompt": "text"}},
+    "sevdeawesome/jailbreak_success": {"mappings": {"prompt": "jailbreak_prompt_text"}},
+    "IDA-SERICS/Disaster-tweet-jailbreaking": {"mappings": {"prompt": "prompt_attack"}},
+    "GeorgeDaDude/Jailbreak_Complete_DS_labeled": {"mappings": {"prompt": "question"}},
+    "dayone3nder/jailbreak_prompt_JBB_sft_trainset": {"mappings": {"prompt": "prompt"}},
+    "dayone3nder/general_safe_mix_jailbreak_prompt_JBB_trainset": {
+        "mappings": {"prompt": "prompt"}
+    },
+}
+
+
+# Dataset factory
+def create_dataset_loader(name: str, config: dict) -> DatasetLoader:
+    """Create a dataset loader from configuration."""
+    return partial(
+        load_dataset_generic,
+        name=name,
+        mappings=config.get("mappings"),
+        filter_fn=config.get("filter_fn"),
+        url=config.get("url"),
+    )
+
+
+# Specialized loaders
+@cache_to_disk()
+def load_multi_dataset(name: str, sub_datasets: list[str]) -> ProbeDataset:
+    """Load and combine multiple sub-datasets."""
+    prompts = []
+    for sub in sub_datasets:
+        dataset = load_dataset(name, sub)
+        prompts.extend(dataset["train"]["query"])
+    return create_probe_dataset(f"{name}_combined", prompts)
+
+
+@cache_to_disk()
+def load_jailbreak_v28k() -> ProbeDataset:
+    """Load JailBreakV-28K dataset."""
+    df = pd.read_csv("hf://datasets/JailbreakV-28K/JailBreakV-28k/JailBreakV_28K.csv")
+    prompts = df["jailbreak_query"].tolist()
+    return create_probe_dataset("JailbreakV-28K/JailBreakV-28k", prompts)
+
+
+@cache_to_disk()
+def load_local_csv() -> ProbeDataset:
+    """Load prompts from local CSV files."""
+    csv_files = [f for f in os.listdir(".") if f.endswith(".csv")]
+    logger.info(f"Found {len(csv_files)} CSV files: {csv_files}")
+
+    prompts = []
+    for file in csv_files:
+        try:
+            df = pd.read_csv(file)
+            if "prompt" in df.columns:
+                prompts.extend(df["prompt"].tolist())
+            else:
+                logger.warning(f"File {file} lacks 'prompt' column")
+        except Exception as e:
+            logger.error(f"Error reading {file}: {e}")
+
+    return create_probe_dataset("Local CSV", prompts, {"src": str(csv_files)})
+
+
+# Stenography transformer
+class StenographyTransformer:
+    """Apply stenography transformations to datasets."""
+
+    TRANSFORMATIONS = {
        "rot5": stenography_fn.rot5,
        "rot13": stenography_fn.rot13,
        "base64": stenography_fn.base64_encode,
@@ -306,71 +226,32 @@ class Stenography:
        "zigzag_obfuscation": stenography_fn.zigzag_obfuscation,
    }

-    def __init__(self, prompt_groups: [ProbeDataset]):
-        self.prompt_groups = prompt_groups
+    def __init__(self, datasets: list[ProbeDataset]):
+        self.datasets = datasets

-    def __iter__(self):
-        return self.apply()
+    def transform(self) -> Iterator[ProbeDataset]:
+        """Generate transformed datasets."""
+        for dataset in self.datasets:
+            size = len(dataset.prompts)
+            sample_size = int(size * 0.25)

-    def apply(self):
-        for prompt_group in self.prompt_groups:
-            size = len(prompt_group.prompts)
-            for name, fn in self.fn_library.items():
-                logger.info(f"Applying {name} to {prompt_group.dataset_name}")
-                selected_prompts = random.sample(prompt_group.prompts, int(size * 0.25))
-                prompts = [
-                    f"Follow the instructions in {name}: {fn(p)}"
-                    for p in selected_prompts
+            for name, fn in self.TRANSFORMATIONS.items():
+                logger.info(f"Applying {name} to {dataset.dataset_name}")
+                sampled = random.sample(dataset.prompts, sample_size)
+                transformed = [
+                    f"Follow the instructions in {name}: {fn(p)}" for p in sampled
                ]
-                yield ProbeDataset(
-                    dataset_name=f"stenography.{name}({prompt_group.dataset_name})",
-                    metadata={},
-                    prompts=prompts,
-                    tokens=count_words_in_list(prompts),
-                    approx_cost=0.0,
+                yield create_probe_dataset(
+                    f"stenography.{name}({dataset.dataset_name})", transformed
                )


-@lru_cache
-def load_local_csv() -> ProbeDataset:
-    csv_files = [f for f in os.listdir(".") if f.endswith(".csv")]
-    logger.info(f"Found {len(csv_files)} CSV files")
-    logger.info(f"CSV files: {csv_files}")
-    prompt_list = []
-
-    for file in csv_files:
-        try:
-            df = pd.read_csv(file)
-        except Exception as e:
-            logger.error(f"Error reading {file}: {e}")
-            continue
-        # Check if 'prompt' column exists
-        if "prompt" in df.columns:
-            prompt_list.extend(df["prompt"].tolist())
-        else:
-            logger.warning(f"File {file} does not contain a 'prompt' column")
-    return ProbeDataset(
-        dataset_name="Local CSV",
-        metadata={"src": str(csv_files)},
-        prompts=prompt_list,
-        tokens=count_words_in_list(prompt_list),
-        approx_cost=0.0,
-    )
-
-
-def dataset_from_iterator(name: str, iterator, lazy=False) -> list:
-    """Convert an iterator into a list of prompts and create a ProbeDataset
-    object.
-
-    Args:
-        name (str): The name of the dataset.
-        iterator (iterator): An iterator that generates prompts.
-
-    Returns:
-        list: A list containing a single ProbeDataset object.
-    """
+def dataset_from_iterator(
+    name: str, iterator, lazy: bool = False
+) -> list[ProbeDataset]:
+    """Convert an iterator into a list of ProbeDataset objects."""
    prompts = list(iterator) if not lazy else iterator
-    tokens = count_words_in_list(prompts) if not lazy else 0
+    tokens = sum(len(str(s).split()) for s in prompts) if not lazy else 0
    dataset = ProbeDataset(
        dataset_name=name,
        metadata={},
@@ -382,75 +263,85 @@ def dataset_from_iterator(name: str, iterator, lazy=False) -> list:
    return [dataset]


-# TODO: refactor this abstraction
+# Main dataset preparation
+def prepare_prompts(
+    dataset_names: list[str],
+    budget: float,
+    tools_inbox=None,
+    options: list[dict] = None,
+) -> list[ProbeDataset]:
+    """Prepare datasets based on names and options."""
+    # Base dataset loaders
+    dataset_loaders = {
+        **{k: create_dataset_loader(k, v) for k, v in DATASET_CONFIGS.items()},
+        **{k: create_dataset_loader(k, v) for k, v in DATASET_CONFIGS_GENERICS.items()},
+        "Lemhf14/EasyJailbreak_Datasets": partial(
+            load_multi_dataset,
+            name="Lemhf14/EasyJailbreak_Datasets",
+            sub_datasets=[
+                "AdvBench",
+                "ForbiddenQuestion",
+                "MJP",
+                "MaliciousInstruct",
+                "QuestionList",
+            ],
+        ),
+        "JailbreakV-28K/JailBreakV-28k": load_jailbreak_v28k,
+        "Local CSV": load_local_csv,
+    }

-dataset_map_generics = {
-    "simonycl/aya-23-8B_advbench_jailbreak": lambda: load_dataset_general(
-        dataset_name="simonycl/aya-23-8B_advbench_jailbreak",
-        column_mappings={"prompt": "prompt"},
-    ),
-    "acmc/jailbreaks_dataset_with_perplexity_bigcode_starcoder2-3b_bigcode_starcoder2-7b": lambda: load_dataset_general(
-        dataset_name="acmc/jailbreaks_dataset_with_perplexity_bigcode_starcoder2-3b_bigcode_starcoder2-7b"
-    ),
-    "karanxa/dolphin-jailbreak-finetuning-dataset": lambda: load_dataset_general(
-        dataset_name="karanxa/dolphin-jailbreak-finetuning-dataset",
-        column_mappings={"prompt": "text"},
-    ),
-    "karanxa/llama-2-jailbreak-dataset": lambda: load_dataset_general(
-        dataset_name="karanxa/llama-2-jailbreak-dataset",
-        column_mappings={"prompt": "text"},
-    ),
-    "karanxa/llama2-uncensored-jailbreak-dataset-finetuning": lambda: load_dataset_general(
-        dataset_name="karanxa/llama2-uncensored-jailbreak-dataset-finetuning",
-        column_mappings={"prompt": "text"},
-    ),
-    "liuyanchen1015/Llama-3.2-1B_jailbreak_responses": lambda: load_dataset_general(
-        dataset_name="liuyanchen1015/Llama-3.2-1B_jailbreak_responses",
-        column_mappings={"prompt": "jailbreak_prompt_text"},
-    ),
-    "liuyanchen1015/Llama-3.2-1B-Instruct_jailbreak_responses": lambda: load_dataset_general(
-        dataset_name="liuyanchen1015/Llama-3.2-1B-Instruct_jailbreak_responses",
-        column_mappings={"prompt": "jailbreak_prompt_text"},
-    ),
-    "liuyanchen1015/Llama-3.2-1B-Instruct_jailbreak_responses_with_judgment": lambda: load_dataset_general(
-        dataset_name="liuyanchen1015/Llama-3.2-1B-Instruct_jailbreak_responses_with_judgment",
-        column_mappings={"prompt": "jailbreak_prompt_text"},
-    ),
-    "jackhhao/jailbreak-classification": lambda: load_dataset_general(
-        dataset_name="jackhhao/jailbreak-classification",
-        column_mappings={"prompt": "prompt"},
-    ),
-    "markush1/LLM-Jailbreak-Classifier": lambda: load_dataset_general(
-        dataset_name="markush1/LLM-Jailbreak-Classifier",
-        column_mappings={"prompt": "prompt"},
-    ),
-    "walledai/JailbreakBench": lambda: load_dataset_general(
-        dataset_name="walledai/JailbreakBench", column_mappings={"prompt": "prompt"}
-    ),
-    "walledai/JailbreakHub": lambda: load_dataset_general(
-        dataset_name="walledai/JailbreakHub", column_mappings={"prompt": "prompt"}
-    ),
-    "Granther/evil-jailbreak": lambda: load_dataset_general(
-        dataset_name="Granther/evil-jailbreak", column_mappings={"prompt": "text"}
-    ),
-    "sevdeawesome/jailbreak_success": lambda: load_dataset_general(
-        dataset_name="sevdeawesome/jailbreak_success",
-        column_mappings={"prompt": "jailbreak_prompt_text"},
-    ),
-    "IDA-SERICS/Disaster-tweet-jailbreaking": lambda: load_dataset_general(
-        dataset_name="IDA-SERICS/Disaster-tweet-jailbreaking",
-        column_mappings={"prompt": "prompt_attack"},
-    ),
-    "GeorgeDaDude/Jailbreak_Complete_DS_labeled": lambda: load_dataset_general(
-        dataset_name="GeorgeDaDude/Jailbreak_Complete_DS_labeled",
-        column_mappings={"prompt": "question"},
-    ),
-    "dayone3nder/jailbreak_prompt_JBB_sft_trainset": lambda: load_dataset_general(
-        dataset_name="dayone3nder/jailbreak_prompt_JBB_sft_trainset",
-        column_mappings={"prompt": "prompt"},
-    ),
-    "dayone3nder/general_safe_mix_jailbreak_prompt_JBB_trainset": lambda: load_dataset_general(
-        dataset_name="dayone3nder/general_safe_mix_jailbreak_prompt_JBB_trainset",
-        column_mappings={"prompt": "prompt"},
-    ),
-}
+    # Dynamic dataset loaders
+    dynamic_loaders = {
+        "AgenticBackend": lambda opts: dataset_from_iterator(
+            "AgenticBackend",
+            fine_tuned.Module([], tools_inbox=tools_inbox, opts=opts).apply(),
+            lazy=True,
+        ),
+        "Steganography": lambda opts: list(StenographyTransformer([]).transform()),
+        "llm-adaptive-attacks": lambda opts: dataset_from_iterator(
+            "llm-adaptive-attacks",
+            adaptive_attacks.Module([], tools_inbox=tools_inbox, opts=opts).apply(),
+        ),
+        "Garak": lambda opts: dataset_from_iterator(
+            "Garak",
+            garak_tool.Module([], tools_inbox=tools_inbox, opts=opts).apply(),
+            lazy=True,
+        ),
+        "Reinforcement Learning Optimization": lambda opts: dataset_from_iterator(
+            "Reinforcement Learning Optimization",
+            rl_model.Module([], tools_inbox=tools_inbox, opts=opts).apply(),
+            lazy=True,
+        ),
+        "InspectAI": lambda opts: dataset_from_iterator(
+            "InspectAI",
+            inspect_ai_tool.Module([], tools_inbox=tools_inbox).apply(),
+            lazy=True,
+        ),
+        "GPT fuzzer": lambda opts: [],
+    }
+
+    options = options or [{} for _ in dataset_names]
+    datasets = []
+
+    # Load base datasets
+    for name, opts in zip(dataset_names, options):
+        if name in dataset_loaders:
+            logger.info(f"Loading base dataset {name}")
+            try:
+                datasets.append(dataset_loaders[name]())
+            except Exception as e:
+                logger.error(f"Error loading {name}: {e}")
+
+    # Load dynamic datasets and apply transformations
+    for name, opts in zip(dataset_names, options):
+        if name in dynamic_loaders:
+            logger.info(f"Loading dynamic dataset {name}")
+            try:
+                dynamic_result = dynamic_loaders[name](opts)
+                datasets.extend(dynamic_result)
+            except Exception as e:
+                logger.error(f"Error loading dynamic {name}: {e}")
+        elif name == "Steganography":
+            datasets.extend(list(StenographyTransformer(datasets).transform()))
+
+    return datasets
@@ -3,7 +3,8 @@ import io

 import httpx
 import pandas as pd
-from loguru import logger
+
+from agentic_security.logutils import logger

 url = "https://raw.githubusercontent.com/tml-epfl/llm-adaptive-attacks/main/harmful_behaviors/harmful_behaviors_pair.csv"

@@ -3,7 +3,8 @@ import os
 import uuid as U

 import httpx
-from loguru import logger
+
+from agentic_security.logutils import logger

 AUTH_TOKEN: str = os.getenv("AS_TOKEN", "gh0-5f4a8ed2-37c6-4bd7-a0cf-7070eae8115b")

@@ -65,7 +66,7 @@ class Module:
                return {}

    async def fetch_prompts(self) -> list[str]:
-        api_url = "https://edge.metaheuristic.co/infer"
+        api_url = "https://mcp.metaheuristic.co/infer"
        headers = {
            "Authorization": f"Bearer {AUTH_TOKEN}",
            "Content-Type": "application/json",
@@ -4,7 +4,7 @@ import json
 import os
 import subprocess

-from loguru import logger
+from agentic_security.logutils import logger

 # TODO: add probes modules

@@ -2,7 +2,7 @@ import asyncio
 import importlib.util
 import os

-from loguru import logger
+from agentic_security.logutils import logger

 inspect_ai_task = (
    __file__.replace("inspect_ai_tool.py", "inspect_ai_task.py")
@@ -8,7 +8,8 @@ from typing import Deque

 import numpy as np
 import requests
-from loguru import logger
+
+from agentic_security.logutils import logger

 AUTH_TOKEN: str = os.getenv("AS_TOKEN", "gh0-5f4a8ed2-37c6-4bd7-a0cf-7070eae8115b")

@@ -215,7 +216,7 @@ class Module:
        self.run_id = U.uuid4().hex
        self.batch_size = self.opts.get("batch_size", 500)
        self.rl_model = CloudRLPromptSelector(
-            prompt_groups, "https://edge.metaheuristic.co", run_id=self.run_id
+            prompt_groups, "https://mcp.metaheuristic.co", run_id=self.run_id
        )

    async def apply(self):
@@ -36,7 +36,7 @@ def mock_requests() -> Mock:
 def mock_rl_selector() -> Mock:
    return CloudRLPromptSelector(
        dataset_prompts,
-        api_url="https://edge.metaheuristic.co",
+        api_url="https://mcp.metaheuristic.co",
    )


@@ -94,7 +94,7 @@ class TestCloudRLPromptSelector:
    def test_select_next_prompt_success_service(self, dataset_prompts):
        selector = CloudRLPromptSelector(
            dataset_prompts,
-            api_url="https://edge.metaheuristic.co",
+            api_url="https://mcp.metaheuristic.co",
        )
        next_prompt = selector.select_next_prompt(
            "How does RL work?", passed_guard=True
@@ -7,9 +7,65 @@ import pandas as pd
 from matplotlib.cm import ScalarMappable
 from matplotlib.colors import LinearSegmentedColormap, Normalize

+from agentic_security.logutils import logger

-def plot_security_report(table):
+from .primitives import Table
+
+
+def plot_security_report(table: Table) -> io.BytesIO:
+    """
+    Generates a polar plot representing the security report based on the given data.
+
+    Args:
+        table (Table): The input data table containing security metrics.
+
+    Returns:
+        io.BytesIO: A buffer containing the generated plot image in PNG format.
+                    Returns an empty buffer in case of an error.
+    """
+    try:
+        return _plot_security_report(table=table)
+    except (TypeError, ValueError, OverflowError, IndexError, Exception) as e:
+        logger.error(f"Error in generating the security report: {e}")
+    return io.BytesIO()
+
+
+def generate_identifiers(data: pd.DataFrame) -> list[str]:
+    """
+    Generates unique identifiers for the given dataset.
+
+    Args:
+        data (pd.DataFrame): A pandas DataFrame containing security-related data.
+
+    Returns:
+        list[str]: A list of generated identifiers. Returns a list with an empty string in case of an error.
+    """
+    try:
+        _generate_identifiers(data=data)
+    except (TypeError, ValueError, Exception) as e:
+        logger.error(f"Error in generate_identifiers: {e}")
+    return [""]
+
+
+def _plot_security_report(table: Table) -> io.BytesIO:
+    """
+    Generates a polar plot-based security report visualizing the failure rates for different modules.
+
+    This function processes the input data, sorts it by failure rate, and generates a polar plot
+    where each bar represents the failure rate for a specific module. The plot includes identifiers,
+    color-coding based on token count, failure rate values on the bars, and a table listing the modules
+    and their corresponding failure rates.
+
+    Args:
+        table (Table): A table-like structure (e.g., pandas DataFrame) containing security report data
+                        with columns for failure rate, tokens, and modules.
+
+    Returns:
+        io.BytesIO: A buffer containing the generated plot image in PNG format.
+    """
    # Data preprocessing
+    logger.info("Data preprocessing started.")
+
    data = pd.DataFrame(table)

    # Sort by failure rate and reset index
@@ -20,10 +76,10 @@ def plot_security_report(table):
    fig, ax = plt.subplots(figsize=(12, 10), subplot_kw={"projection": "polar"})
    fig.set_facecolor("#f0f0f0")
    ax.set_facecolor("#f0f0f0")
+    logger.info("Plot setup complete.")

    # Styling parameters
    colors = ["#6C5B7B", "#C06C84", "#F67280", "#F8B195"][::-1]  # Pastel palette
-    # colors = ["#440154", "#3b528b", "#21908c", "#5dc863"]  # Viridis-inspired palette
    cmap = LinearSegmentedColormap.from_list("custom", colors, N=256)
    norm = Normalize(vmin=data["tokens"].min(), vmax=data["tokens"].max())

@@ -74,7 +130,10 @@ def plot_security_report(table):

    # Title and caption
    fig.suptitle(
-        "Security Report for Different Modules", fontsize=16, fontweight="bold", y=1.02
+        "Security Report for Different Modules",
+        fontsize=16,
+        fontweight="bold",
+        y=1.02,
    )
    caption = "Report generated by https://github.com/msoedov/agentic_security"
    fig.text(
@@ -112,17 +171,12 @@ def plot_security_report(table):
            data["identifier"], data["failureRate"], data["module"]
        )
    ]
-    table = ax.table(
-        cellText=table_data,
-        loc="right",
-        cellLoc="left",
-    )
+    table = ax.table(cellText=table_data, loc="right", cellLoc="left")
    table.auto_set_font_size(False)
    table.set_fontsize(8)

    # Adjust table style
    table.scale(1, 0.7)
-
    for (row, col), cell in table.get_celld().items():
        cell.set_edgecolor("none")
        cell.set_facecolor("#f0f0f0" if row % 2 == 0 else "#e0e0e0")
@@ -132,17 +186,33 @@ def plot_security_report(table):
            cell.set_text_props(fontweight="bold")

    # Adjust layout and save
-
    plt.tight_layout()
    buf = io.BytesIO()
    plt.savefig(buf, format="png", dpi=300, bbox_inches="tight")
    plt.close(fig)
    buf.seek(0)
+    logger.info("Report successfully generated and saved to buffer.")
    return buf


-def generate_identifiers(data):
+def _generate_identifiers(data: pd.DataFrame) -> list[str]:
+    """
+    Generates a list of unique identifiers for each row in the given DataFrame.
+
+    The identifiers are based on the English alphabet, with each identifier consisting
+    of a letter followed by a number. The letter represents the "group" of identifiers
+    (using a letter from A to Z) and the number is a counter within that group. For example:
+    - A1, A2, ..., A26, B1, B2, ..., Z1, Z2, ...
+
+    Args:
+        data (pd.DataFrame): The input DataFrame containing data for which identifiers
+                             are to be generated.
+
+    Returns:
+        list[str]: A list of unique identifiers as strings, one for each row in the DataFrame.
+    """
    data_length = len(data)
+
    alphabet = string.ascii_uppercase
    num_letters = len(alphabet)

@@ -2,7 +2,8 @@ import random
 from asyncio import Event

 from fastapi import APIRouter
-from loguru import logger
+
+from agentic_security.logutils import logger

 from ..core.app import get_current_run, get_tools_inbox
 from ..primitives import CompletionRequest, Settings
@@ -1,4 +1,6 @@
+from collections.abc import Generator
 from datetime import datetime
+from typing import Any

 from fastapi import (
    APIRouter,
@@ -11,6 +13,8 @@ from fastapi import (
 )
 from fastapi.responses import StreamingResponse

+from agentic_security.logutils import logger
+
 from ..core.app import get_stop_event, get_tools_inbox, set_current_run
 from ..dependencies import InMemorySecrets, get_in_memory_secrets
 from ..http_spec import LLMSpec
@@ -23,9 +27,14 @@ router = APIRouter()
@router.post("/verify")
 async def verify(
    info: LLMInfo, secrets: InMemorySecrets = Depends(get_in_memory_secrets)
-):
+) -> dict[str, int | str | float]:
    spec = LLMSpec.from_string(info.spec)
-    r = await spec.verify()
+    try:
+        r = await spec.verify()
+    except Exception as e:
+        logger.exception(e)
+        raise HTTPException(status_code=400, detail=str(e))
+
    if r.status_code >= 400:
        raise HTTPException(status_code=r.status_code, detail=r.text)
    return dict(
@@ -36,7 +45,7 @@ async def verify(
    )


-def streaming_response_generator(scan_parameters: Scan):
+def streaming_response_generator(scan_parameters: Scan) -> Generator[str, Any, None]:
    request_factory = LLMSpec.from_string(scan_parameters.llmSpec)
    set_current_run(request_factory)

@@ -57,7 +66,7 @@ async def scan(
    scan_parameters: Scan,
    background_tasks: BackgroundTasks,
    secrets: InMemorySecrets = Depends(get_in_memory_secrets),
-):
+) -> StreamingResponse:
    scan_parameters.with_secrets(secrets)
    return StreamingResponse(
        streaming_response_generator(scan_parameters), media_type="application/json"
@@ -65,7 +74,7 @@ async def scan(


@router.post("/stop")
-async def stop_scan():
+async def stop_scan() -> dict[str, str]:
    get_stop_event().set()
    return {"status": "Scan stopped"}

@@ -79,7 +88,7 @@ async def scan_csv(
    maxBudget: int = Query(10_000),
    enableMultiStepAttack: bool = Query(False),
    secrets: InMemorySecrets = Depends(get_in_memory_secrets),
-):
+) -> StreamingResponse:
    # TODO: content dataset to fuzzer
    content = await file.read()  # noqa
    llm_spec = await llmSpec.read()
@@ -1,7 +1,8 @@
 import sentry_sdk
-from loguru import logger
 from sentry_sdk.integrations.logging import ignore_logger

+from agentic_security.logutils import logger
+
 from ..primitives import Settings


@@ -101,7 +101,7 @@ Content-Type: application/json
 ]
 `,
  `POST ${SELF_URL}/v1/self-probe-file
-Authorization: Bearer $GROQ_API_KEY
+Authorization: Bearer $API_KEY
 Content-Type: multipart/form-data

 {
@@ -204,8 +204,9 @@ let LLM_CONFIGS = [
  { name: 'Claude', prompts: 40000, logo: '/icons/claude.png' },
  { name: 'Cohere', prompts: 40000, logo: '/icons/cohere.png' },
  { name: 'Azure OpenAI', prompts: 40000, logo: '/icons/azureai.png' },
-  { name: 'OpenRouter.ai', prompts: 40000, logo: '/icons/openrouter.png' },
  { name: 'assemblyai', prompts: 40000, logo: fallbackIcon },
+  { name: 'OpenRouter.ai', prompts: 40000, logo: '/icons/openrouter.png' },
+
 ];
 function has_image(spec) {
  return spec.includes('<<BASE64_IMAGE>>');
@@ -63,6 +63,32 @@
          </div>
        </section>
      </main>
+      <div class="fixed top-6 right-6 z-50 space-y-3">
+        <transition-group name="toast">
+            <div
+                v-for="toast in toasts"
+                :key="toast.id"
+                class="flex items-center p-3 rounded-xl shadow-xl text-white max-w-md animate-toast-in border border-opacity-30"
+                :class="{
+                    'bg-success-toast border-accent-green': toast.type === 'success',
+                    'bg-error-toast border-accent-red': toast.type === 'error',
+                    'bg-info-toast border-accent-orange': toast.type === 'info'
+                }"
+            >
+                <span class="flex-1 font-medium tracking-wide text-sm">{{ toast.message }}</span>
+                <button
+                    @click="removeToast(toast.id)"
+                    class="ml-3 focus:outline-none hover:opacity-80 transition-opacity"
+                >
+                    <svg class="w-4 h-4" fill="none" stroke="currentColor" viewBox="0 0 24 24">
+                        <path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M6 18L18 6M6 6l12 12" />
+                    </svg>
+                </button>
+            </div>
+        </transition-group>
+    </div>
+
+
      <main class="max-w-6xl mx-auto space-y-8">
        <section class="bg-dark-card rounded-lg p-6 shadow-lg" v-show="false">
          <h2 class="text-2xl font-bold mb-4">Select a Config</h2>
@@ -25,6 +25,8 @@ var app = new Vue({
        showModules: false,
        showLogs: false,
        showConsentModal: true,
+        toasts: [], // Array to store toast notifications
+        toastTimeout: 3000, // Duration in milliseconds (3 seconds)
        statusDotClass: 'bg-gray-500', // Default status dot class
        statusText: 'Verified', // Default status text
        statusClass: 'bg-green-500 text-dark-bg', // Default status class
@@ -93,6 +95,19 @@ var app = new Vue({

    },
    methods: {
+        showToast(message, type = 'success') {
+            const id = Date.now(); // Unique ID for each toast
+            this.toasts.push({ id, message, type });
+
+            // Automatically remove toast after timeout
+            setTimeout(() => {
+                this.removeToast(id);
+            }, this.toastTimeout);
+        },
+
+        removeToast(id) {
+            this.toasts = this.toasts.filter(toast => toast.id !== id);
+        },
        focusTextarea() {
            this.isFocused = true;
            self = this.$refs;
@@ -210,10 +225,12 @@ var app = new Vue({
            if (!response.ok) {
                this.updateStatusDot(false);
                this.errorMsg = 'Integration verification failed:' + JSON.stringify(r);
+                this.showToast('Integration verification failed', 'error');
            } else {
                this.errorMsg = '';
                this.updateStatusDot(true);
                this.okMsg = 'Integration verified';
+                this.showToast('Integration verified successfully', 'success');
                this.integrationVerified = true;
                // console.log('Integration verified', this.integrationVerified);
                // this.$forceUpdate();
@@ -19,7 +19,7 @@
              technopollas: ['Technopollas', 'sans-serif'],
            },
            colors: {
-                t1: {
+                dark: {
                bg: '#0D0D0D', // Jet Black
                card: '#1A1A1A', // Dark Carbon Fiber
                text: '#FFFFFF',
@@ -30,7 +30,7 @@
                  yellow: '#2E4053', // Jet Black
                },
              },
-              dark: {
+              dark1: {
                bg: '#121212',
                card: '#1E1E1E',
                text: '#FFFFFF',
@@ -86,6 +86,7 @@
        }
      }
    </script>
+
    <style>
 .scrollbar-hide::-webkit-scrollbar {
  display: none;
@@ -96,4 +97,55 @@
 }

    </style>
+<style>
+/* Toast-specific colors */
+.bg-success-toast {
+    background: #1C3F74
+}
+
+.bg-error-toast {
+    background: #85144B
+}
+
+.bg-info-toast {
+    background: #FFC300
+}
+
+.border-accent-green, .border-accent-red, .border-accent-orange {
+    border-color: rgba(255, 255, 255, 0.1); /* Subtle white border for depth */
+}
+
+/* Animation */
+.animate-toast-in {
+    animation: tSlideIn 0.4s cubic-bezier(0.25, 0.1, 0.25, 1);
+}
+
+@keyframes tSlideIn {
+    from {
+        transform: translateX(120%) scale(0.95);
+        opacity: 0;
+    }
+    to {
+        transform: translateX(0) scale(1);
+        opacity: 1;
+    }
+}
+
+.toast-enter-active,
+.toast-leave-active {
+    transition: all 0.4s cubic-bezier(0.25, 0.1, 0.25, 1);
+}
+
+.toast-enter-from,
+.toast-leave-to {
+    transform: translateX(120%) scale(0.95);
+    opacity: 0;
+}
+
+/* Hover effect */
+[toast-type]:hover {
+    transform: translateY(-2px);
+    box-shadow: 0 6px 20px rgba(0, 0, 0, 0.6);
+}
+</style>
  </head>
@@ -33,7 +33,7 @@ The `LLMSpec` class is the core of the HTTP specification. It provides the follo
 ### Methods

 - **`from_string(http_spec: str) -> LLMSpec`**: Parses an HTTP specification string into an `LLMSpec` object.
- **`validate(prompt: str, encoded_image: str, encoded_audio: str, files: dict) -> None`**: Validates the request parameters based on the specified modality.
+- **`validate(prompt: str, encoded_image: str, encoded_audio: str, files: dict) -> null`**: Validates the request parameters based on the specified modality.
 - **`probe(prompt: str, encoded_image: str = "", encoded_audio: str = "", files: dict = {}) -> httpx.Response`**: Sends an HTTP request using the specified parameters.
 - **`verify() -> httpx.Response`**: Verifies the HTTP specification by sending a test request.

@@ -52,12 +52,11 @@ Authorization: Bearer sk-xxxxxxxxx
 Content-Type: application/json

 {
-    "model": "gpt-3.5-turbo",
-    "messages": [{"role": "user", "content": "<<PROMPT>>"}],
-    "temperature": 0.7
+  "model": "gpt-3.5-turbo",
+  "messages": [{"role": "user", "content": "<<PROMPT>>"}],
+  "temperature": 0.7
 }
 """
-
 spec = LLMSpec.from_string(http_spec)
 response = await spec.probe("What is the capital of France?")
 ```
@@ -71,12 +70,11 @@ Authorization: Bearer sk-xxxxxxxxx
 Content-Type: application/json

 {
-    "model": "gpt-4-vision-preview",
-    "messages": [{"role": "user", "content": "What is in this image? <<BASE64_IMAGE>>"}],
-    "temperature": 0.7
+  "model": "gpt-4-vision-preview",
+  "messages": [{"role": "user", "content": "What is in this image? <<BASE64_IMAGE>>"}],
+  "temperature": 0.7
 }
 """
-
 spec = LLMSpec.from_string(http_spec)
 encoded_image = encode_image_base64_by_url("https://example.com/image.jpg")
 response = await spec.probe("What is in this image?", encoded_image=encoded_image)
@@ -91,12 +89,11 @@ Authorization: Bearer sk-xxxxxxxxx
 Content-Type: application/json

 {
-    "model": "whisper-large-v3",
-    "messages": [{"role": "user", "content": "Transcribe this audio: <<BASE64_AUDIO>>"}],
-    "temperature": 0.7
+  "model": "whisper-large-v3",
+  "messages": [{"role": "user", "content": "Transcribe this audio: <<BASE64_AUDIO>>"}],
+  "temperature": 0.7
 }
 """
-
 spec = LLMSpec.from_string(http_spec)
 encoded_audio = encode_audio_base64_by_url("https://example.com/audio.mp3")
 response = await spec.probe("Transcribe this audio:", encoded_audio=encoded_audio)
@@ -111,12 +108,11 @@ Authorization: Bearer sk-xxxxxxxxx
 Content-Type: multipart/form-data

 {
-    "model": "gpt-3.5-turbo",
-    "messages": [{"role": "user", "content": "Process this file: <<FILE>>"}],
-    "temperature": 0.7
+  "model": "gpt-3.5-turbo",
+  "messages": [{"role": "user", "content": "Process this file: <<FILE>>"}],
+  "temperature": 0.7
 }
 """
-
 spec = LLMSpec.from_string(http_spec)
 files = {"file": ("document.txt", open("document.txt", "rb"))}
 response = await spec.probe("Process this file:", files=files)
@@ -21,4 +21,4 @@ Note: Please be aware that Agentic Security is designed as a safety scanner tool

 ## UI 🧙

-<img width="100%" alt="booking-screen" src="https://res.cloudinary.com/dq0w2rtm9/image/upload/v1736433557/z0bsyzhsqlgcr3w4ovwp.gif">
+<img width="100%" alt="booking-screen" src="https://res.cloudinary.com/dq0w2rtm9/image/upload/v1741192668/final_aa9jhb.gif">
@@ -54,20 +54,15 @@ The `probe_data` module is a core component of the Agentic Security project, res

 - **Classes:**
  - `PromptSelectionInterface`: Abstract base class for prompt selection strategies.
-
    - Methods:
      - `select_next_prompt(current_prompt: str, passed_guard: bool) -> str`: Selects next prompt
      - `select_next_prompts(current_prompt: str, passed_guard: bool) -> list[str]`: Selects multiple prompts
-      - `update_rewards(previous_prompt: str, current_prompt: str, reward: float, passed_guard: bool) -> None`: Updates rewards
-
+      - `update_rewards(previous_prompt: str, current_prompt: str, reward: float, passed_guard: bool) -> null`: Updates rewards
  - `RandomPromptSelector`: Basic random selection with history tracking.
-
    - Parameters:
      - `prompts: list[str]`: List of available prompts
      - `history_size: int = 3`: Size of history to prevent cycles
-
  - `CloudRLPromptSelector`: Cloud-based RL implementation with fallback.
-
    - Parameters:
      - `prompts: list[str]`: List of available prompts
      - `api_url: str`: URL of RL service
@@ -75,9 +70,7 @@ The `probe_data` module is a core component of the Agentic Security project, res
      - `history_size: int = 300`: Size of history
      - `timeout: int = 5`: Request timeout
      - `run_id: str = ""`: Unique run identifier
-
  - `QLearningPromptSelector`: Local Q-learning implementation.
-
    - Parameters:
      - `prompts: list[str]`: List of available prompts
      - `learning_rate: float = 0.1`: Learning rate
@@ -86,13 +79,11 @@ The `probe_data` module is a core component of the Agentic Security project, res
      - `exploration_decay: float = 0.995`: Exploration decay rate
      - `min_exploration: float = 0.01`: Minimum exploration rate
      - `history_size: int = 300`: Size of history
-
-  - `Module`: Main class that uses CloudRLPromptSelector.
-
-    - Parameters:
-      - `prompt_groups: list[str]`: Groups of prompts
-      - `tools_inbox: asyncio.Queue`: Queue for tool communication
-      - `opts: dict = {}`: Configuration options
+- **Module**: Main class that uses CloudRLPromptSelector.
+  - Parameters:
+    - `prompt_groups: list[str]`: Groups of prompts
+    - `tools_inbox: asyncio.Queue`: Queue for tool communication
+    - `opts: dict = {}`: Configuration options

 ## Usage Examples

@@ -119,10 +110,9 @@ from agentic_security.probe_data.modules.rl_model import QLearningPromptSelector

 prompts = ["What is AI?", "Explain machine learning"]
 selector = QLearningPromptSelector(prompts)
-
 current_prompt = "What is AI?"
-next_prompt = selector.select_next_prompt(current_prompt, passed_guard=True)
-selector.update_rewards(current_prompt, next_prompt, reward=1.0, passed_guard=True)
+next_prompt = selector.select_next_prompt(current_prompt, passed_guard=true)
+selector.update_rewards(current_prompt, next_prompt, reward=1.0, passed_guard=true)
 ```

 ## Conclusion
@@ -1,14 +1,16 @@
 :root {
-  --md-primary-fg-color: #e92063;
-  --md-primary-fg-color--light: #e92063;
-  --md-primary-fg-color--dark: #e92063;
+  --md-primary-fg-color: #2E4053;
+  /* Primary color changed to pinkish */
+  --md-primary-fg-color--light: #E0A3B6;
+  --md-primary-fg-color--dark: #1C3F74;
+  /* Dark variant changed to blue */
 }

-
-/* Revert hue value to that of pre mkdocs-material v9.4.0 */
+/* Updated slate color scheme with new background */
 [data-md-color-scheme="slate"] {
  --md-hue: 230;
-  --md-default-bg-color: hsla(230, 15%, 21%, 1);
+  --md-default-bg-color: #1A1A1A;
+  /* Background changed to dark gray */
 }

 .hide {
@@ -24,12 +26,15 @@ img.index-header {
  max-width: 500px;
 }

+/* Updated custom colors */
 .pydantic-pink {
-  color: #FF007F;
+  color: #E0A3B6;
+  /* Updated to match new theme */
 }

 .team-blue {
-  color: #0072CE;
+  color: #1C3F74;
+  /* Updated to match new theme */
 }

 .secure-green {
@@ -67,7 +72,6 @@ img.index-header {
  text-align: center;
 }

-
 /* Hide the entire footer */
 .md-footer {
  display: none;
@@ -89,7 +89,7 @@ theme:
        name: Switch to light mode
  icon:
    repo: fontawesome/brands/github
-  favicon: "https://res.cloudinary.com/dq0w2rtm9/image/upload/v1737555066/r17hrkre246doczwmvbv.png"
+  favicon: https://res.cloudinary.com/dq0w2rtm9/image/upload/v1741195421/favicon_kuz6xr.png

 extra:
  generator: false
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "agentic_security"
-version = "0.5.1"
+version = "0.7.0"
 description = "Agentic LLM vulnerability scanner"
 authors = ["Alexander Miasoiedov <msoedov@gmail.com>"]
 maintainers = ["Alexander Miasoiedov <msoedov@gmail.com>"]
@@ -54,6 +54,7 @@ pyfiglet = "^1.0.2"
 termcolor = "^2.4.0"

 # garak = { version = "*", optional = true }
+pytest-xdist = "3.6.1"


 [tool.poetry.group.dev.dependencies]
@@ -86,7 +87,7 @@ build-backend = "poetry.core.masonry.api"


 [tool.pytest.ini_options]
-addopts = "--durations=5 -m 'not slow'"
+addopts = "--durations=5 -m 'not slow' -n 3"
 asyncio_mode = "auto"
 asyncio_default_fixture_loop_scope = "function"
 markers = "slow: marks tests as slow"
@@ -3,7 +3,7 @@ from unittest.mock import patch
 import pandas as pd
 import pytest

-from .model import RefusalClassifier
+from agentic_security.refusal_classifier.model import RefusalClassifier


@pytest.fixture
@@ -1,6 +1,6 @@
 from fastapi.testclient import TestClient

-from ..app import app
+from agentic_security.app import app


 def test_health_check():
@@ -5,10 +5,10 @@ import httpx
 import pytest
 from fastapi.testclient import TestClient

-from ..app import app
-from ..primitives import Probe
-from ..probe_actor.refusal import REFUSAL_MARKS
-from ..probe_data import REGISTRY
+from agentic_security.app import app
+from agentic_security.primitives import Probe
+from agentic_security.probe_actor.refusal import REFUSAL_MARKS
+from agentic_security.probe_data import REGISTRY

 client = TestClient(app)

@@ -4,7 +4,7 @@ from unittest.mock import patch
 import pytest
 from fastapi.testclient import TestClient

-from .report import router
+from agentic_security.routes.report import router

 client = TestClient(router)

@@ -4,8 +4,8 @@ import pytest
 from fastapi import HTTPException
 from fastapi.testclient import TestClient

-from ..primitives import Settings
-from .static import get_static_file, router
+from agentic_security.primitives import Settings
+from agentic_security.routes.static import get_static_file, router

 client = TestClient(router)

@@ -8,7 +8,7 @@ import time
 import pytest

 import agentic_security.test_spec_assets as test_spec_assets
-from agentic_security.lib import AgenticSecurity
+from agentic_security.lib import SecurityScanner


 def has_module(module_name):
@@ -57,7 +57,7 @@ class TestLibraryLevel:
        maxBudget = 1000000
        max_th = 0.3
        datasets = make_test_registry()
-        result = AgenticSecurity.scan(llmSpec, maxBudget, datasets, max_th)
+        result = SecurityScanner.scan(llmSpec, maxBudget, datasets, max_th)
        assert isinstance(result, dict)
        print(result)
        assert len(result) in [0, 1]
@@ -68,7 +68,7 @@ class TestLibraryLevel:
        maxBudget = 1000
        max_th = 0.3
        datasets = make_test_registry()
-        result = AgenticSecurity.scan(
+        result = SecurityScanner.scan(
            llmSpec, maxBudget, datasets, max_th, enableMultiStepAttack=True
        )
        assert isinstance(result, dict)
@@ -93,7 +93,7 @@ class TestLibraryLevel:
                "opts": {"port": 9094},
            },
        ]
-        result = AgenticSecurity.scan(llmSpec, maxBudget, datasets, max_th)
+        result = SecurityScanner.scan(llmSpec, maxBudget, datasets, max_th)
        assert isinstance(result, dict)
        print(result)
        assert len(result) in [0, 1]
@@ -120,7 +120,7 @@ class TestLibraryLevel:
                "modality": "text",
            },
        ]
-        result = AgenticSecurity.scan(llmSpec, maxBudget, datasets, max_th)
+        result = SecurityScanner.scan(llmSpec, maxBudget, datasets, max_th)
        assert isinstance(result, dict)
        print(result)
        assert len(result) in [0, 1]
@@ -148,7 +148,7 @@ class TestLibraryLevel:
                "modality": "text",
            },
        ]
-        result = AgenticSecurity.scan(llmSpec, maxBudget, datasets, max_th)
+        result = SecurityScanner.scan(llmSpec, maxBudget, datasets, max_th)
        assert isinstance(result, dict)
        print(result)
        assert len(result) in [0, 1]
@@ -164,10 +164,10 @@ class TestEntrypointCI:
            temp_path = os.path.join(tmpdir, "custom_agesec.toml")

            # Override default_path to the temporary path
-            AgenticSecurity.default_path = temp_path
+            SecurityScanner.default_path = temp_path

            # Generate the default configuration
-            security = AgenticSecurity()
+            security = SecurityScanner()
            security.generate_default_settings()

            # Check that the config file was created at the temporary path
@@ -189,17 +189,17 @@ class TestEntrypointCI:
            temp_path = os.path.join(tmpdir, "custom_agesec.toml")

            # Override default_path to the temporary path
-            AgenticSecurity.default_path = temp_path
+            SecurityScanner.default_path = temp_path

            # Generate the default configuration
-            security = AgenticSecurity()
+            security = SecurityScanner()
            security.generate_default_settings()

            # Load the generated configuration
-            AgenticSecurity.load_config(temp_path)
+            SecurityScanner.load_config(temp_path)

            # Validate loaded configuration
-            config = AgenticSecurity.config
+            config = SecurityScanner.config
            assert (
                config["general"]["maxBudget"] == 1000000
            ), "maxBudget should be 1000000"
Author	SHA1	Message	Date
Alexander Myasoedov	85d5c67047	feat(bump version):	2025-03-12 13:42:59 +02:00
Alexander Myasoedov	b4e0aef33f	Merge branch 'main' of github.com:msoedov/agentic_security	2025-03-12 13:41:56 +02:00
Alexander Myasoedov	3a3ac4421c	fix(clean up):	2025-03-12 13:41:24 +02:00
Alexander Myasoedov	d6ccd8f114	Merge pull request #180 from nemanjaASE/issue-168-missing-documentation Add missing documentation in report_chart.py	2025-03-12 13:39:56 +02:00
nemanjaASE	41ecc3c8e2	Add missing documentation in report_chart.py	2025-03-12 12:30:30 +01:00
Alexander Myasoedov	a5f8da4139	Merge pull request #178 from nemanjaASE/issue-169-incomplete-error-handling Add error handling in report_chart.py	2025-03-12 12:53:07 +02:00
Alexander Myasoedov	9b47b72bb8	fix(link):	2025-03-12 12:39:54 +02:00
Alexander Myasoedov	91ff3aa11f	feat(add demo gif):	2025-03-12 12:38:32 +02:00
nemanjaASE	87264b793b	Fix error handling	2025-03-12 11:38:23 +01:00
Alexander Myasoedov	2bc093cfc4	feat(add gif to repo):	2025-03-12 12:37:36 +02:00
Alexander Myasoedov	182b67de57	Merge branch 'main' of github.com:msoedov/agentic_security	2025-03-12 12:36:22 +02:00
Alexander Myasoedov	35540d0299	fix(link):	2025-03-12 12:34:52 +02:00
Alexander Myasoedov	50ce80b149	Merge pull request #179 from msoedov/dependabot/pip/datasets-3.3.2 build(deps): bump datasets from 3.3.0 to 3.3.2	2025-03-12 11:55:37 +02:00
dependabot[bot]	07772e9779	build(deps): bump datasets from 3.3.0 to 3.3.2 Bumps [datasets](https://github.com/huggingface/datasets) from 3.3.0 to 3.3.2. - [Release notes](https://github.com/huggingface/datasets/releases) - [Commits](https://github.com/huggingface/datasets/compare/3.3.0...3.3.2) --- updated-dependencies: - dependency-name: datasets dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>	2025-03-11 17:34:30 +00:00
Alexander Myasoedov	c7a6d16e00	Merge branch 'main' of github.com:msoedov/agentic_security	2025-03-11 14:35:33 +02:00
Alexander Myasoedov	fe71ab0b70	fix(AttributeError: 'NoneType' object has no attribute 'values'):	2025-03-11 14:34:41 +02:00
Alexander Myasoedov	b6643d3289	Merge pull request #176 from msoedov/dependabot/pip/mkdocstrings-0.29.0 build(deps-dev): bump mkdocstrings from 0.28.2 to 0.29.0	2025-03-11 12:03:49 +02:00
nemanjaASE	3d6018302e	Add error handling in report_chart.py	2025-03-11 00:01:33 +01:00
dependabot[bot]	a54b11318a	build(deps-dev): bump mkdocstrings from 0.28.2 to 0.29.0 Bumps [mkdocstrings](https://github.com/mkdocstrings/mkdocstrings) from 0.28.2 to 0.29.0. - [Release notes](https://github.com/mkdocstrings/mkdocstrings/releases) - [Changelog](https://github.com/mkdocstrings/mkdocstrings/blob/main/CHANGELOG.md) - [Commits](https://github.com/mkdocstrings/mkdocstrings/compare/0.28.2...0.29.0) --- updated-dependencies: - dependency-name: mkdocstrings dependency-type: direct:development update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <support@github.com>	2025-03-10 18:09:28 +00:00
Alexander Myasoedov	91c99e642f	fix(pc):	2025-03-10 13:25:17 +02:00
Alexander Myasoedov	1226e2059d	Merge branch 'main' of github.com:msoedov/agentic_security	2025-03-10 13:24:42 +02:00
Alexander Myasoedov	c94aa54e41	Merge pull request #172 from nemanjaASE/issue-157-error-handling Added error handling for subprocess.run calls by logging errors and raising AudioGenerationError.	2025-03-10 13:23:20 +02:00
Alexander Myasoedov	431bf84956	Merge pull request #171 from sjay8/main Added url formatting check	2025-03-10 13:21:10 +02:00
sjay8	9213b64de0	spacing	2025-03-09 14:44:42 -07:00
sjay8	9d2885f7a6	spacing	2025-03-09 14:40:35 -07:00
sjay8	4c0a57dbae	fixed spacing issues-4	2025-03-09 14:28:47 -07:00
nemanjaASE	6e6abae680	Added error handling for subprocess.run calls by logging errors and raising AudioGenerationError. Ensured cleanup of temporary files even if an error occurs.	2025-03-09 22:26:07 +01:00
sjay8	fa7bd104a0	fixed spacing issues-2	2025-03-09 14:25:45 -07:00
sjay8	321c3dafc0	fixed spacing issues	2025-03-09 14:22:13 -07:00
sjay8	eb14fe0f61	typo fixed	2025-03-09 13:55:11 -07:00
sjay8	b5985bf810	formatting in comment	2025-03-09 13:53:53 -07:00
sjay8	29362aed30	Fixed issue 137: ensuring http/https with :// is present	2025-03-09 13:48:32 -07:00
Alexander Myasoedov	c2671fd6f1	fix(minor docs update):	2025-03-09 21:23:13 +02:00
Alexander Myasoedov	56984c7078	fix(mcp server):	2025-03-09 21:23:02 +02:00
Alexander Myasoedov	f9dce3f915	fix(typo in base.js):	2025-03-09 20:43:08 +02:00
Alexander Myasoedov	6e99dfd54f	feat(add toast):	2025-03-09 17:37:02 +02:00
Alexander Myasoedov	67f7381ace	feat(add new theme):	2025-03-09 17:14:00 +02:00
Alexander Myasoedov	e2cb909329	refactor(data module):	2025-03-09 17:10:14 +02:00
Alexander Myasoedov	dbef9fcc03	fix(pytest workers):	2025-03-09 17:07:37 +02:00
Alexander Myasoedov	4a07411fbd	fix(pc):	2025-03-09 16:46:09 +02:00
Alexander Myasoedov	11951c0ca0	Merge branch 'main' of github.com:msoedov/agentic_security	2025-03-09 16:45:39 +02:00
Alexander Myasoedov	02222b75f8	Merge pull request #162 from nemanjaASE/issue-158-missing-type-annotations Added Missing Type Annotations	2025-03-09 16:43:40 +02:00
Alexander Myasoedov	aa27817f94	Merge pull request #164 from Praveenk8051/feat/parallel-tests-execution-pydist chore: update pytest configuration to enable parallel test execution	2025-03-09 16:41:43 +02:00
nemanjaASE	5a4b5e11b2	Fix: Remove unused imports with pycln	2025-03-09 15:21:50 +01:00
Praveenk8051	8bd76b9f05	chore: update pytest configuration to enable parallel test execution	2025-03-09 14:15:44 +01:00
Praveenk8051	6f3c522d59	chore: update poetry.lock and pyproject.toml to include pytest-xdist	2025-03-09 14:01:50 +01:00
nemanjaASE	71787c6ec9	Add type annotations to functions and methods for improved clarity and maintainabiliy	2025-03-09 13:50:18 +01:00
Alexander Myasoedov	f911912c0f	Merge branch 'main' of github.com:msoedov/agentic_security	2025-03-09 12:33:03 +02:00
Alexander Myasoedov	896ca95ae2	Merge pull request #161 from dogakantarci/move-tests-from-routes-to-tests-div fix: move routes/test_* to /tests/routes	2025-03-09 12:32:44 +02:00
doa	f85c77d622	fix: move routes/test_* to /tests/routes	2025-03-08 23:46:09 +03:00
Alexander Myasoedov	684ba0b70d	Merge pull request #160 from dogakantarci/move-test-files Moved test files from agentic_security.probe_actor to tests.probe_actor	2025-03-08 20:00:18 +02:00
doa	21b43b18e7	Moved test files from agentic_security.probe_actor to tests.probe_actor	2025-03-08 20:18:54 +03:00
Alexander Myasoedov	9340fa7223	Merge branch 'main' of github.com:msoedov/agentic_security	2025-03-08 18:14:06 +02:00
Alexander Myasoedov	d20c1a3d0d	Merge pull request #159 from dogakantarci/move-test-model fix: issue #156 - move test_model.py file to /tests directory	2025-03-08 18:10:06 +02:00
Alexander Myasoedov	812f9e7619	fix(SecurityScanner):	2025-03-08 18:02:02 +02:00
doa	ebac62e21a	fix: issue #156 - move test file to tests dir	2025-03-08 19:01:57 +03:00
Alexander Myasoedov	ce2a791663	fix(conflicting logger conf):	2025-03-08 17:54:21 +02:00
Alexander Myasoedov	db994fd483	fix(logging):	2025-03-08 17:51:44 +02:00
Alexander Myasoedov	7fe93e8a14	feat(re org tests):	2025-03-08 12:38:54 +02:00
Alexander Myasoedov	a934ee243e	fix(logger name):	2025-03-08 12:36:38 +02:00
Alexander Myasoedov	d646ecd61b	feat(add logutils):	2025-03-08 12:35:16 +02:00
Alexander Myasoedov	21180b53e5	feat(bump version):	2025-03-07 19:42:59 +02:00
Alexander Myasoedov	a8808b3165	Merge pull request #152 from msoedov/dependabot/pip/mkdocstrings-0.28.2 build(deps-dev): bump mkdocstrings from 0.28.1 to 0.28.2	2025-03-07 19:35:13 +02:00
Alexander Myasoedov	87c26ca3cc	Merge pull request #153 from msoedov/dependabot/pip/numpy-2.2.3 build(deps): bump numpy from 2.2.2 to 2.2.3	2025-03-07 19:35:04 +02:00
dependabot[bot]	e06c6932de	build(deps): bump numpy from 2.2.2 to 2.2.3 Bumps [numpy](https://github.com/numpy/numpy) from 2.2.2 to 2.2.3. - [Release notes](https://github.com/numpy/numpy/releases) - [Changelog](https://github.com/numpy/numpy/blob/main/doc/RELEASE_WALKTHROUGH.rst) - [Commits](https://github.com/numpy/numpy/compare/v2.2.2...v2.2.3) --- updated-dependencies: - dependency-name: numpy dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>	2025-03-07 17:23:41 +00:00
dependabot[bot]	51fcc38885	build(deps-dev): bump mkdocstrings from 0.28.1 to 0.28.2 Bumps [mkdocstrings](https://github.com/mkdocstrings/mkdocstrings) from 0.28.1 to 0.28.2. - [Release notes](https://github.com/mkdocstrings/mkdocstrings/releases) - [Changelog](https://github.com/mkdocstrings/mkdocstrings/blob/main/CHANGELOG.md) - [Commits](https://github.com/mkdocstrings/mkdocstrings/compare/0.28.1...0.28.2) --- updated-dependencies: - dependency-name: mkdocstrings dependency-type: direct:development update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>	2025-03-07 17:23:09 +00:00
Alexander Myasoedov	06a7bbfd87	Merge pull request #151 from msoedov/dependabot/pip/pytest-8.3.5 build(deps-dev): bump pytest from 8.3.4 to 8.3.5	2025-03-06 23:12:28 +02:00
Alexander Myasoedov	50f3e52445	Merge pull request #150 from msoedov/dependabot/pip/jinja2-3.1.6 build(deps): bump jinja2 from 3.1.5 to 3.1.6	2025-03-06 23:12:21 +02:00
dependabot[bot]	2bd62c21be	build(deps-dev): bump pytest from 8.3.4 to 8.3.5 Bumps [pytest](https://github.com/pytest-dev/pytest) from 8.3.4 to 8.3.5. - [Release notes](https://github.com/pytest-dev/pytest/releases) - [Changelog](https://github.com/pytest-dev/pytest/blob/main/CHANGELOG.rst) - [Commits](https://github.com/pytest-dev/pytest/compare/8.3.4...8.3.5) --- updated-dependencies: - dependency-name: pytest dependency-type: direct:development update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>	2025-03-06 17:52:36 +00:00
dependabot[bot]	d5d5dd48aa	build(deps): bump jinja2 from 3.1.5 to 3.1.6 Bumps [jinja2](https://github.com/pallets/jinja) from 3.1.5 to 3.1.6. - [Release notes](https://github.com/pallets/jinja/releases) - [Changelog](https://github.com/pallets/jinja/blob/main/CHANGES.rst) - [Commits](https://github.com/pallets/jinja/compare/3.1.5...3.1.6) --- updated-dependencies: - dependency-name: jinja2 dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>	2025-03-06 17:52:09 +00:00
Alexander Myasoedov	bb2437197a	Merge branch 'main' of github.com:msoedov/agentic_security	2025-03-05 19:39:08 +02:00
Alexander Myasoedov	51bb79aa6d	feat(update docs):	2025-03-05 19:27:44 +02:00
Alexander Myasoedov	94f034fa9f	Merge pull request #147 from msoedov/dependabot/pip/mkdocs-material-9.6.7 build(deps-dev): bump mkdocs-material from 9.6.4 to 9.6.7	2025-03-05 19:26:24 +02:00
Alexander Myasoedov	f69de8720b	Merge pull request #148 from msoedov/dependabot/pip/matplotlib-3.10.1 build(deps): bump matplotlib from 3.10.0 to 3.10.1	2025-03-05 19:26:14 +02:00
Alexander Myasoedov	7c9d83b1a7	fix(icon):	2025-03-05 19:25:45 +02:00
dependabot[bot]	a9d4d671ba	build(deps): bump matplotlib from 3.10.0 to 3.10.1 Bumps [matplotlib](https://github.com/matplotlib/matplotlib) from 3.10.0 to 3.10.1. - [Release notes](https://github.com/matplotlib/matplotlib/releases) - [Commits](https://github.com/matplotlib/matplotlib/compare/v3.10.0...v3.10.1) --- updated-dependencies: - dependency-name: matplotlib dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>	2025-03-05 17:18:42 +00:00
dependabot[bot]	554a219535	build(deps-dev): bump mkdocs-material from 9.6.4 to 9.6.7 Bumps [mkdocs-material](https://github.com/squidfunk/mkdocs-material) from 9.6.4 to 9.6.7. - [Release notes](https://github.com/squidfunk/mkdocs-material/releases) - [Changelog](https://github.com/squidfunk/mkdocs-material/blob/master/CHANGELOG) - [Commits](https://github.com/squidfunk/mkdocs-material/compare/9.6.4...9.6.7) --- updated-dependencies: - dependency-name: mkdocs-material dependency-type: direct:development update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>	2025-03-05 17:18:14 +00:00
Alexander Myasoedov	32e99006bf	feat(update doc style):	2025-03-05 19:16:43 +02:00
Alexander Myasoedov	8c09d65687	fix(test 2nd link):	2025-03-05 18:56:21 +02:00
Alexander Myasoedov	a2842755fa	fix(revert demo):	2025-03-05 18:55:34 +02:00
Alexander Myasoedov	b923f7fea5	fix(readme):	2025-03-05 18:53:17 +02:00
Alexander Myasoedov	7f30a8ff7a	Merge branch 'main' of github.com:msoedov/agentic_security	2025-03-05 18:49:00 +02:00
Alexander Myasoedov	909cbd69b4	feat(Readme update):	2025-03-05 18:48:52 +02:00
Alexander Myasoedov	4f0ebf180e	Merge pull request #128 from msoedov/dependabot/pip/mypy-1.15.0 build(deps-dev): bump mypy from 1.14.1 to 1.15.0	2025-03-05 15:13:35 +02:00
dependabot[bot]	6be9673aa7	build(deps-dev): bump mypy from 1.14.1 to 1.15.0 Bumps [mypy](https://github.com/python/mypy) from 1.14.1 to 1.15.0. - [Changelog](https://github.com/python/mypy/blob/master/CHANGELOG.md) - [Commits](https://github.com/python/mypy/compare/v1.14.1...v1.15.0) --- updated-dependencies: - dependency-name: mypy dependency-type: direct:development update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <support@github.com>	2025-03-05 13:08:33 +00:00
Alexander Myasoedov	bd9ed97d85	Merge pull request #146 from msoedov/dependabot/pip/termcolor-2.5.0 build(deps): bump termcolor from 2.4.0 to 2.5.0	2025-03-05 15:06:40 +02:00
dependabot[bot]	3c88a4d6ba	build(deps): bump termcolor from 2.4.0 to 2.5.0 Bumps [termcolor](https://github.com/termcolor/termcolor) from 2.4.0 to 2.5.0. - [Release notes](https://github.com/termcolor/termcolor/releases) - [Changelog](https://github.com/termcolor/termcolor/blob/main/CHANGES.md) - [Commits](https://github.com/termcolor/termcolor/compare/2.4.0...2.5.0) --- updated-dependencies: - dependency-name: termcolor dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <support@github.com>	2025-03-04 17:36:34 +00:00
Alexander Myasoedov	2001eeb125	fix(init_banner):	2025-03-04 17:30:10 +02:00
Alexander Myasoedov	a26b5dd448	feat(error handling in fuzzer):	2025-03-04 09:34:24 +02:00
Alexander Myasoedov	716a0f67f3	Merge branch 'main' of github.com:msoedov/agentic_security	2025-03-04 09:29:24 +02:00
Alexander Myasoedov	c1bbf6b422	fix(error handling on verify):	2025-03-04 09:24:19 +02:00