feat: add timeout to Agent.generate_response()

[rosspeili]

Problem Statement

If a local model (e.g., Ollama) or an external API provider hangs or becomes unresponsive, the session is currently blocked indefinitely. There is no timeout mechanism in the Agent.generate_response() call, which can lead to a "frozen" interface.

Proposed Solution

1. Add a timeout: int = Field(default=30) field to the AgentConfig model in config.py.
2. Pass this timeout parameter to the litellm.completion() call within agent.py.
3. Wrap the completion call in a try/except block to catch litellm.Timeout errors.
4. Return a graceful error message (or trigger a retry) instead of allowing the process to hang.

Alternatives Considered

None. Robustness against infrastructure failure is a core requirement.

Priority

High 🔴

Additional Context

Essential for reliability in local environments where local model serving might be unstable or slow.