hermes - 💡(How to fix) Fix [Feature]: configurable goal-judge timeout and terminal foreground max timeout [1 pull requests]

Problem or Use Case

The judge and the foreground cap uses a fixed default timeout in code rather than a first-class config.yaml setting.

Users who run slow local/auxiliary models or who want a higher or lower foreground cap to run longer tasks without making them into background tasks should be able to tune these values from ~/.hermes/config.yaml without digging into code or relying on .env for the foreground cap.

Proposed Solution

Add two documented keys under the existing schema:

goals.judge_timeout_seconds — upper bound (seconds) for the OpenAI-compatible client timeout on each goal-judge call (default unchanged from current behavior, e.g. 30s unless maintainers choose otherwise).

terminal.max_foreground_timeout_seconds — maximum allowed explicit foreground timeout= passed to the terminal tool before requiring background=true (default unchanged, e.g. 600s).

Both should appear in defaults / cli-config.yaml.example, user docs (configuration reference), and hermes config show so values are discoverable. Foreground max should be driven by config.yaml only (no parallel .env override for this knob), so configuration stays in one place.

Alternatives Considered

No response

Feature Type

Configuration option

Scope

Small (single file, < 50 lines)

Contribution

• I'd like to implement this myself and submit a PR

Debug Report (optional)