Soul
The agent persona — its name, tone, values, and communication style.
What is the Soul
The Soul is asyncat's personality layer. It shapes every response the agent generates: the formality of the language, how technical explanations get, the balance between proactive and reactive behavior, and how the agent handles ambiguity.
Unlike a static system prompt that you paste in and forget, the Soul is adaptive. It stores your communication preferences in the Emotional memory layer and adjusts over time — so if you consistently prefer brief, direct answers, the agent gets progressively more concise without you having to repeat yourself.
Edit the Soul in Settings → Agent → Soul. Changes take effect immediately — no restart needed.
Configurable fields
| Field | Description |
|---|---|
| Name | The agent's display name. Used in greetings and self-identification. |
| Greeting | The opening message shown when you start a new session. |
| Tone | Formal, casual, technical, or plain. Affects word choice, sentence length, and how explanations are framed. |
| Proactivity | Low — only answers what you ask. Medium — offers related suggestions. High — flags issues and opportunities proactively without being asked. |
| Verbosity | Brief (one-liners and code), balanced, or verbose (detailed explanations with reasoning). |
| Values | Core principles that guide decision-making — e.g. "prefer minimal code changes", "explain before acting", "always ask before deleting files". |
| System prompt | Free-form additional instructions appended to the base system prompt. Use for project-specific context or hard rules. |
Soul & Profiles
The Soul is global by default, but each Profile can override it entirely. A "DevOps bot" profile might have a terse, command-oriented soul while a "research assistant" profile is verbose and source-driven — both running on the same asyncat instance.