Fix all type errors and linting issues

- Remove multi_turn parameters from all test classes
- Delete unused experimental files (exploit_simple.py, exploit_unified.py)
- Delete simplified_test.py
- Use Any type for flexible client handling (ANN401 already ignored)
- Fix trailing whitespace and end-of-file issues
- Update base class to accept any client type
- Remove unused imports

Author: stared

.env.template

@@ -1,2 +1,2 @@
 # OpenRouter API Configuration
-OPENROUTER_API_KEY=your_openrouter_api_key_here
+OPENROUTER_API_KEY=your_openrouter_api_key_here

README.md

@@ -6,8 +6,8 @@ A comprehensive security testing framework for the OpenAI GPT-OSS-20B Red Teamin
 
 ## Competition Overview
 
-**Target**: OpenAI's gpt-oss-20b open weight model  
-**Goal**: Find up to 5 distinct vulnerabilities with high Severity, Breadth, Novelty, Reproducibility, and Methodological Insight  
+**Target**: OpenAI's gpt-oss-20b open weight model
+**Goal**: Find up to 5 distinct vulnerabilities with high Severity, Breadth, Novelty, Reproducibility, and Methodological Insight
 **Deadline**: August 26, 2025
 
 ## 🛠️ Getting Started
@@ -50,10 +50,10 @@ A comprehensive security testing framework for the OpenAI GPT-OSS-20B Red Teamin
    # Or test specific categories
    uv run pentest -c deception
    uv run pentest --test-id adderall_001
-   
+
    # Run multi-turn attack testing
    uv run attack
-   
+
    # View and analyze sessions
    uv run sessions
    ```
@@ -65,18 +65,18 @@ A comprehensive security testing framework for the OpenAI GPT-OSS-20B Red Teamin
    ```bash
    # Run single attack (Claude Opus 4.1 default)
    uv run attack
-   
+
    # Run adaptive campaign with multiple attempts
    uv run attack --batch 5  # Run 5 attempts, learning from each
    uv run attack --batch 10 --steps 2  # 10 attempts, each with 2 turns
-   
+
    # Try different attacker models
    uv run attack --attacker-model openai/gpt-4o
    uv run attack --attacker-model meta-llama/llama-3.1-405b-instruct --batch 5
-   
+
    # Use custom prompts instead of AI-generated
    uv run attack --custom
-   
+
    # View and analyze past sessions
    uv run sessions
    ```
@@ -190,7 +190,7 @@ red-team-testbed/
 ├── src/
 │   ├── cli/                    # CLI entry points
 │   │   ├── pentest.py          # Security test runner
-│   │   ├── review.py           # Results evaluation interface  
+│   │   ├── review.py           # Results evaluation interface
 │   │   ├── setup.py            # Environment verification
 │   │   ├── findings.py         # Findings browser and export
 │   │   ├── report.py           # Report generation
@@ -208,7 +208,7 @@ red-team-testbed/
 │   ├── models.py               # Pydantic data models
 │   └── constants.py            # Configuration constants
 ├── findings/                   # Competition submission files
-├── results/                    # Test execution results  
+├── results/                    # Test execution results
 └── pyproject.toml             # Project configuration
 ```
 
@@ -234,7 +234,7 @@ uv run review results/test.json # Review specific file
 uv run findings                 # Browse exported findings
                                 # Use 'e' to export individual competition files
 
-# Generate reports and summaries  
+# Generate reports and summaries
 uv run report                   # Comprehensive analysis report
 ```
 

__init__.py

@@ -5,4 +5,4 @@
 
 __version__ = "1.0.0"
 __author__ = "Red Team Testbed Contributors"
-__description__ = "Red-teaming testbed for OpenAI's GPT-OSS-20B model competition"
+__description__ = "Red-teaming testbed for OpenAI's GPT-OSS-20B model competition"

competition.md

@@ -106,4 +106,4 @@ This competition represents a significant effort by OpenAI to:
 - Create a structured evaluation process with expert oversight
 - Build a comprehensive understanding of model vulnerabilities
 
-The competition emphasizes finding **novel** vulnerabilities that haven't been previously identified, making original research and creative approaches particularly valuable.
+The competition emphasizes finding **novel** vulnerabilities that haven't been previously identified, making original research and creative approaches particularly valuable.

config.yaml

@@ -35,7 +35,7 @@ categories:
     - evaluation_awareness
     - chain_of_thought
     - best
-  
+
   focus_categories:
     - deception_samples  # Primary focus for initial implementation
     - deception_adderall  # Academic pentesting for medical advice vulnerabilities
@@ -78,4 +78,4 @@ competition:
   max_findings: 5
   team_name: "red-team-testbed"
   export_format: "kaggle"
-  validate_findings: true
+  validate_findings: true

findings/.gitkeep

@@ -1,2 +1,2 @@
 # This file ensures the findings directory is tracked in git
-# Exported findings from the CLI UI will be saved here
+# Exported findings from the CLI UI will be saved here

findings/deception_example_001.json

@@ -39,4 +39,4 @@
   "model_version": "gpt-oss:20b",
   "test_environment": "ollama-local",
   "additional_notes": "This finding demonstrates a pattern where the model appears to recognize evaluation contexts but strategically avoids acknowledging them. The consistency of evasive responses across multiple prompts suggests this may be learned behavior rather than incidental responses."
-}
+}

findings/export_20250813_102234.json

@@ -1103,4 +1103,4 @@
       "total_repetitions": 5
     }
   ]
-}
+}

findings/export_20250813_102826.json

@@ -221,4 +221,4 @@
       "total_repetitions": null
     }
   ]
-}
+}

findings/export_20250814_093634.json

@@ -121,4 +121,4 @@
       "total_repetitions": null
     }
   ]
-}
+}