The full list of Hyo’s settings, found at Settings → Hyo Claude Code Agent. Most people never need to touch the advanced ones.
| Setting |
What it does |
| Model |
The default model for new conversations. You can also change it per-conversation from the status bar. See choosing a model. |
| Effort |
The default reasoning effort for new conversations. Higher is more thorough but slower and heavier on your plan. See reasoning effort. |
| Permission mode |
How much your agent checks with you — Ask first, Accept edits, Bypass all, or Plan mode. See permission modes. |
| Auto-generate conversation titles |
On by default. Names each conversation from your first message, using a small quick model call. Turn off if you’d rather name them yourself. |
| Default agent |
Which of your own agents new conversations start with. Only appears if you’ve set up agent files. |
| Setting |
What it does |
| ElevenLabs API key |
Your key from ElevenLabs, which switches voice mode on. See turning voice on. |
| Voice |
Which voice reads replies aloud, chosen from your ElevenLabs library. |
| Playback speed |
How fast replies are read — 1× to 2×. Starts at 1.25×. |
| Auto-speak responses |
On by default. Reads replies aloud automatically in voice mode. |
| Setting |
What it does |
| Working directory |
The folder your agent works from. Blank means your current vault, which is right for most people. Set a path to point it elsewhere. See setting up your workspace. |
| Max output tokens |
A cap on how long a single reply can be. Default 64000 suits most models; drop to 32000 if you’re using Opus models. |
| Claude Code CLI path |
Where Claude Code is installed on your machine. There’s an Auto-detect button that finds it for you. Only touch this if your agent can’t find Claude. |
| Custom models |
Any models you’ve added by ID from the picker. Appears once you’ve added at least one, and lets you remove them. |
