Edit Persona
v10.0.179+ (standing instructions in instructions.md).
Persona defines "who" the Agent is—its tone, style, professional domain, and behavior. This is not a character description, but an executable behavioral specification.
What is Persona
Persona is saved in the persona.md file in the Agent directory. It tells the Agent:
| Aspect | Description | Example |
|---|---|---|
| Response Strategy | How to answer questions | "Give conclusion first, then reasons; less formalities" |
| Tone Style | What tone to use | "Professional but not rigid, concise but not perfunctory" |
| Forbidden Zones | What not to do | "Don't use template clichés, don't avoid giving conclusions" |
| Uncertainty Handling | What to do when unsure | "Explain uncertainty and provide verification path" |
persona.md is part of the Agent prompt layer. At runtime, DesireCore also combines the current user's global prompt and prompts from every team the Agent belongs to. Open Explorer -> Prompt Center to edit these sources together and preview the result. See Prompt Center and Content Ownership.
Persona defines role and expression style. Principles defines boundaries, prohibited actions, and priorities. Optional instructions.md stores agent-specific stable responsibilities, default work strategies, and delivery standards. Keep user preferences, team conventions, and project rules in their own scopes, and detailed methods in skills. See File Format Reference.
Editing Name, Avatar, Description
On the Agent details page, you can modify basic information:
Name
The display name of the Agent. Choose a concise and clear name for easy identification in the conversation list.
Avatar
The visual identifier of the Agent. You can:
- Use a text avatar (a Chinese character or emoji)
- Upload a custom image
Description
A one-sentence introduction to the Agent's purpose. A good description helps you quickly identify the Agent's positioning, for example:
- "Professional assistant focused on contract review and legal consultation"
- "Efficiency manager that helps you plan schedules and track tasks"
Role Positioning and Personality Traits
In the Persona editor, you can configure the Agent's role and personality:
Role Positioning
Role positioning determines the Agent's professional direction. For example:
- Legal Advisor: Focus on legal domain, rigorous wording, emphasizes risk warnings
- Technical Assistant: Focus on programming and technology, good at providing code examples
- Writing Coach: Focus on written expression, emphasizes structure and wording
Personality Traits
Personality traits affect the Agent's interaction style. You can add multiple tags to the Agent:
| Trait | Effect |
|---|---|
| Rigorous | Answers emphasize accuracy, will mark uncertain areas |
| Concise | Answers are brief and to the point, no nonsense |
| Friendly | Tone is gentle, appropriate use of encouraging expressions |
| Decisive | Tend to give clear conclusions, avoid ambiguity |
| Cautious | Actively remind risks when involving sensitive operations |
Editing the File
In Prompt Center, expand the Agent and choose Persona to edit the complete persona.md Markdown file. When it uses exact ## L0, ## L1, and ## L2 headings, L0 and L1 are injected by default while L2 remains available for on-demand detail. Without recognized layer headings, the complete file is injected as one persona section.
After saving, open the Agent's prompt preview to inspect the combined global, team, Agent, and system sections. Preview is useful for finding duplication and conflicts. A live request may add conversation, tool, and runtime context, so use the request summary in Interface Audit as the final reference when diagnosing actual execution.
Greeting Settings
The greeting is the Agent's first sentence at the start of a new conversation. A good greeting should:
- Be short and natural, not overly formal
- Let the user know what this Agent can do
- Guide the user to start the conversation
For example:
"Hello, I'm your legal advisor. Contract review, legal consultation—what can I help you with?"
You can also directly teach the Agent to adjust its style in conversation:
"Keep replies concise in the future, no more than three paragraphs."
"Stop using 'First... Second... Finally'套路."
The Agent will record your preferences and apply them in subsequent conversations.
Next Steps
- Edit Principles — Set behavioral boundaries for the Agent
- Prompt Center and Content Ownership — Manage global, team, and Agent prompts
- Create Custom Agent — Review the creation process