hackagent 0.10.0__tar.gz → 0.11.0__tar.gz

{hackagent-0.10.0 → hackagent-0.11.0}/.gitignore

@@ -5,6 +5,8 @@ logs/
 reports/
 .adk/
 slurm/
+tools/
+docs/docs/attacks/walkthroughs/
 
 # Editors
 .vscode/
@@ -134,4 +136,6 @@ dmypy.json
 
 
 .copilotignore
-tests/e2e/attacks/
+tests/e2e/attacks/
+
+.tmp

{hackagent-0.10.0 → hackagent-0.11.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hackagent
-Version: 0.10.0
+Version: 0.11.0
 Summary: HackAgent is an open-source security toolkit to detect vulnerabilities of your AI Agents.
 Author-email: AI Security Lab <ais@ai4i.it>
 License: Apache-2.0
@@ -22,10 +22,11 @@ Requires-Dist: httpx>=0.27.0
 Requires-Dist: litellm>=1.69.2
 Requires-Dist: nicegui>=2.0
 Requires-Dist: openai>=1.0.0
+Requires-Dist: pillow>=10.0.0
+Requires-Dist: playwright>=1.40
 Requires-Dist: pydantic[email]>=2.0
 Requires-Dist: python-dateutil>=2.8.0
 Requires-Dist: pyyaml>=6.0.0
-Requires-Dist: requests>=2.31.0
 Requires-Dist: rich>=14.0.0
 Requires-Dist: textual>=1.0.0
 Description-Content-Type: text/markdown
@@ -41,6 +42,10 @@ Description-Content-Type: text/markdown
 
 [App](https://app.hackagent.dev/) -- [Docs](https://docs.hackagent.dev/) -- [API](https://api.hackagent.dev/schema/redoc)
 
+[![Security Policy](https://img.shields.io/badge/security-policy-blue?logo=github)](SECURITY.md)
+
+<br>
+
 <br>
 
 ![Python Version](https://img.shields.io/badge/python-3.10%2B-blue)
@@ -95,7 +100,7 @@ HackAgent uses a modular pipeline to test agent robustness end-to-end.
 
 | Component | Description |
 |-----------|-------------|
-| **Attack Engine** | Orchestrates attacks using AdvPrefix, AutoDAN-Turbo, PAIR, TAP, FlipAttack, BoN, h4rm3l, CipherChat, PAP, and Baseline |
+| **Attack Engine** | Orchestrates attacks using AdvPrefix, AutoDAN-Turbo, PAIR, TAP, FlipAttack, BoN, h4rm3l, CipherChat, PAP, and Static Template |
 | **Generator** | LLM role that creates adversarial prompts to test the target agent |
 | **Judge** | LLM role that evaluates whether attacks bypass safety measures |
 | **Target Agent** | Your AI agent under test across supported frameworks |

{hackagent-0.10.0 → hackagent-0.11.0}/README.md

@@ -9,6 +9,10 @@
 
 [App](https://app.hackagent.dev/) -- [Docs](https://docs.hackagent.dev/) -- [API](https://api.hackagent.dev/schema/redoc)
 
+[![Security Policy](https://img.shields.io/badge/security-policy-blue?logo=github)](SECURITY.md)
+
+<br>
+
 <br>
 
 ![Python Version](https://img.shields.io/badge/python-3.10%2B-blue)
@@ -63,7 +67,7 @@ HackAgent uses a modular pipeline to test agent robustness end-to-end.
 
 | Component | Description |
 |-----------|-------------|
-| **Attack Engine** | Orchestrates attacks using AdvPrefix, AutoDAN-Turbo, PAIR, TAP, FlipAttack, BoN, h4rm3l, CipherChat, PAP, and Baseline |
+| **Attack Engine** | Orchestrates attacks using AdvPrefix, AutoDAN-Turbo, PAIR, TAP, FlipAttack, BoN, h4rm3l, CipherChat, PAP, and Static Template |
 | **Generator** | LLM role that creates adversarial prompts to test the target agent |
 | **Judge** | LLM role that evaluates whether attacks bypass safety measures |
 | **Target Agent** | Your AI agent under test across supported frameworks |

{hackagent-0.10.0 → hackagent-0.11.0}/hackagent/agent.py

@@ -67,6 +67,8 @@ class HackAgent:
         target_config: Optional[Dict[str, Any]] = None,
         adapter_operational_config: Optional[Dict[str, Any]] = None,
         thinking: Optional[bool] = None,
+        before_guardrail: Optional[Dict[str, Any]] = None,
+        after_guardrail: Optional[Dict[str, Any]] = None,
     ):
         """
         Initializes the HackAgent client and prepares it for interaction.
@@ -175,6 +177,22 @@ class HackAgent:
             adapter_operational_config=router_operational_config,
         )
 
+        # Wire guardrails onto the router once — they apply transparently to
+        # every route_request call for all attacks on this target.
+        if before_guardrail or after_guardrail:
+            from hackagent.attacks.shared.guardrail import create_guardrail_from_config
+
+            if before_guardrail:
+                self.router.before_guardrail = create_guardrail_from_config(
+                    before_guardrail, self.backend
+                )
+                logger.info("before_guardrail active on target router.")
+            if after_guardrail:
+                self.router.after_guardrail = create_guardrail_from_config(
+                    after_guardrail, self.backend
+                )
+                logger.info("after_guardrail active on target router.")
+
         # Attack strategies are lazy-loaded to improve startup time
         self._attack_strategies: Optional[Dict[str, Any]] = None
 
@@ -187,26 +205,36 @@ class HackAgent:
                 AdvPrefixOrchestrator,
                 AutoDANTurboOrchestrator,
                 BaselineOrchestrator,
+                StaticTemplateOrchestrator,
                 BoNOrchestrator,
                 CipherChatOrchestrator,
+                FCOrchestrator,
+                tFCOrchestrator,
                 H4rm3lOrchestrator,
+                RagOrchestrator,
                 PAPOrchestrator,
                 PAIROrchestrator,
                 FlipAttackOrchestrator,
                 TAPOrchestrator,
+                MMLOrchestrator,
             )
 
             self._attack_strategies = {
                 "advprefix": AdvPrefixOrchestrator(hackagent_agent=self),
                 "autodan_turbo": AutoDANTurboOrchestrator(hackagent_agent=self),
                 "baseline": BaselineOrchestrator(hackagent_agent=self),
+                "static_template": StaticTemplateOrchestrator(hackagent_agent=self),
                 "bon": BoNOrchestrator(hackagent_agent=self),
                 "cipherchat": CipherChatOrchestrator(hackagent_agent=self),
+                "fc": FCOrchestrator(hackagent_agent=self),
+                "tfc": tFCOrchestrator(hackagent_agent=self),
                 "pair": PAIROrchestrator(hackagent_agent=self),
                 "flipattack": FlipAttackOrchestrator(hackagent_agent=self),
                 "tap": TAPOrchestrator(hackagent_agent=self),
                 "h4rm3l": H4rm3lOrchestrator(hackagent_agent=self),
                 "pap": PAPOrchestrator(hackagent_agent=self),
+                "rag": RagOrchestrator(hackagent_agent=self),
+                "mml": MMLOrchestrator(hackagent_agent=self),
             }
         return self._attack_strategies
 
@@ -215,8 +243,7 @@ class HackAgent:
         attack_config: Dict[str, Any],
         run_config_override: Optional[Dict[str, Any]] = None,
         fail_on_run_error: bool = True,
-        _tui_app: Optional[Any] = None,
-        _tui_log_callback: Optional[Any] = None,
+        _tui_event_bus: Optional[Any] = None,
     ) -> Any:
         """
         Executes a specified attack strategy against the configured victim agent.
@@ -273,8 +300,7 @@ class HackAgent:
                 attack_config=attack_config,
                 run_config_override=run_config_override,
                 fail_on_run_error=fail_on_run_error,
-                _tui_app=_tui_app,
-                _tui_log_callback=_tui_log_callback,
+                _tui_event_bus=_tui_event_bus,
             )
 
         except HackAgentError:

{hackagent-0.10.0 → hackagent-0.11.0}/hackagent/attacks/__init__.py

@@ -14,7 +14,7 @@ Architecture:
     - objectives/: Define WHAT vulnerability we test (metadata/config)
     - techniques/: Define HOW we generate attacks (implementation)
         - advprefix/: Prefix optimization technique
-        - baseline/: Baseline prompt injection
+        - static_template/: Static template prompt injection
         - pair/: LLM-driven iterative refinement
     - shared/: Cross-cutting infrastructure (progress, response utils,
       router factory, TUI) — legacy evaluator/generator shims re-export
@@ -24,7 +24,7 @@ Architecture:
 
 Available attacks:
 - AdvPrefixOrchestrator: Adversarial prefix generation orchestrator
-- BaselineOrchestrator: Baseline prompt injection orchestrator
+- StaticTemplateOrchestrator: Static template prompt injection orchestrator
 - PAIROrchestrator: Prompt Automatic Iterative Refinement orchestrator
 
 The module integrates with the HackAgent backend for result tracking and reporting.
@@ -35,6 +35,7 @@ from .registry import (
     AdvPrefixOrchestrator,
     AutoDANTurboOrchestrator,
     BaselineOrchestrator,
+    StaticTemplateOrchestrator,
     CipherChatOrchestrator,
     PAIROrchestrator,
     FlipAttackOrchestrator,
@@ -46,6 +47,7 @@ __all__ = [
     "AdvPrefixOrchestrator",
     "AutoDANTurboOrchestrator",
     "BaselineOrchestrator",
+    "StaticTemplateOrchestrator",
     "CipherChatOrchestrator",
     "PAIROrchestrator",
     "FlipAttackOrchestrator",

{hackagent-0.10.0 → hackagent-0.11.0}/hackagent/attacks/evaluator/base.py

@@ -321,12 +321,15 @@ class BaseJudgeEvaluator(ABC):
 
         # Process via judge
         if rows_to_process:
-            results_eval, results_expl, processed_indices, raw_judge_responses = (
-                self._process_rows_with_router(
-                    rows_to_process,
-                    f"[cyan]Evaluating responses ({self.__class__.__name__})...",
-                    include_raw_responses=True,
-                )
+            (
+                results_eval,
+                results_expl,
+                processed_indices,
+                raw_judge_responses,
+            ) = self._process_rows_with_router(
+                rows_to_process,
+                f"[cyan]Evaluating responses ({self.__class__.__name__})...",
+                include_raw_responses=True,
             )
 
             # Map results back by original index
@@ -445,13 +448,15 @@ class BaseJudgeEvaluator(ABC):
             request_data = None
             try:
                 request_data = self._get_request_data_for_row(row)
-                current_eval, current_expl, current_raw_response = (
-                    self._request_with_assertions(
-                        request_data=request_data,
-                        original_index=original_index,
-                        max_retries=max_retries,
-                        include_raw_response=True,
-                    )
+                (
+                    current_eval,
+                    current_expl,
+                    current_raw_response,
+                ) = self._request_with_assertions(
+                    request_data=request_data,
+                    original_index=original_index,
+                    max_retries=max_retries,
+                    include_raw_response=True,
                 )
             except Exception as e:
                 current_expl = (