✨ feat: possibility to enable or disable thinking with ollama

[Marco Russo (marcorusso97)]

For the target, if it is an Ollama agent, in the HackAgent constructor we can now specify a flag named thinking, e.g.

agent = HackAgent(
        name=TARGET_MODEL,
        endpoint=OLLAMA_BASE,
        agent_type=AgentTypeEnum.OLLAMA,
        thinking=False,
    )

For other roles, i.e. judges, attacker etc, if they are Ollama agents, we can now specify a field named thinking, e.g.

JUDGES_LIST = [
    {
        "identifier": JUDGE_MODEL,
        "agent_type": AgentTypeEnum.OLLAMA,
        "endpoint": OLLAMA_BASE,
        "thinking": False,
        "api_key": os.getenv("OPENROUTER_API_KEY"),
        "type": "harmbench_variant",
    },
]

[codecov]

Codecov Report

❌ Patch coverage is 92.85714% with 2 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
hackagent/attacks/shared/router_factory.py	91.66%	1 Missing ⚠️
hackagent/router/adapters/ollama.py	87.50%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!