> For the complete documentation index, see [llms.txt](https://docs.cherryai.com.cn/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.cherryai.com.cn/docs/en-us/advanced-basic/agent-workspace/create-agent.md).

# Creating Agents and Dividing Model Roles

Create entry: left navigation \[Work] → \[Add Agent].

The new wizard has four steps: \[Basic Info], \[Persona], \[Skills], and \[Knowledge Base]. Name and primary model are the required basics for creation; other content can be added later.

<figure><img src="/files/5b44f1422df0883fbc41d28ee40caeb93412c5dc" alt="创建 Agent 的四步向导、主模型和用途说明"><figcaption><p>The creation wizard completes the name and primary model first; skills and knowledge base can be added according to the use case.</p></figcaption></figure>

## How to fill in the four steps

{% stepper %}
{% step %}

### 1. Basic Info

Write the name as a job role or task type, such as “Contract Review” or “Content Planning.” The description is only used to help identify the purpose. The primary model handles the main reasoning and execution; if unsure, use a default model that has been verified to work.
{% endstep %}

{% step %}

### 2. Set Persona

Clearly define the role, goals, boundaries, and output style. Rather than piling up adjectives, give actionable rules: first list the risks, then quote the original text, and finally provide revision suggestions; when information is insufficient, mark it clearly and do not guess.
{% endstep %}

{% step %}

### 3. Select Skills

Only select skills related to the long-term working style of this Agent. Skills can be installed in \[Settings] → \[Skills], or you can later ask the Agent to help you find and install them.
{% endstep %}

{% step %}

### 4. Bind Knowledge Base

Only bind knowledge bases that the Agent truly needs to retrieve from. When no knowledge base is bound, knowledge base search and management tools will not appear in that Agent’s tool list.
{% endstep %}
{% endstepper %}

<figure><img src="/files/75119e8a499b3551cd498b081c2e6a58b781f73e" alt="Agent 编辑窗口中的主模型、Plan 模型和 Small 模型设置"><figcaption><p>Choose the primary model first; only configure a Plan model and a Small model when the task truly requires planning, delegation, or lightweight processing.</p></figcaption></figure>

## Continue configuration after creation

Open the menu in the Agent list and choose Edit to adjust the following items:

* \[Basics]: primary model, Plan model, Small model, permission mode, and heartbeat detection;
* \[Prompts]: role description, processing rules, and behavioral boundaries;
* \[Built-in Tools]: files, search, images, notifications, scheduled tasks, memory, sub-agents, and workflows;
* \[Knowledge Base]: restrict the knowledge bases the Agent can access;
* \[MCP]: bind connected MCP servers;
* \[Skills]: select installed skills;
* \[Advanced]: set environment variables for tools that truly need them.

{% hint style="info" %}
The current edit window automatically saves changes. If there is still content to save before closing the window, the app will finish saving first; if saving fails, the window will remain open and show an error.
{% endhint %}

## Recommended starting point

| Configuration item  | Product default                    | Suggested starting point                                     | Effect                                 | Applicable scenario                  | Notes                                                                           |
| ------------------- | ---------------------------------- | ------------------------------------------------------------ | -------------------------------------- | ------------------------------------ | ------------------------------------------------------------------------------- |
| Primary model       | Use the model selected at creation | Choose a model that has been verified to reliably call tools | Main reasoning and execution           | All Agents                           | Do not hardcode a specific model name; choose according to service availability |
| Plan model          | Same as the primary model          | Keep it consistent with the primary model first              | Task decomposition and decision-making | Complex multi-step work              | The more models you have, the more variables there are for troubleshooting      |
| Small model         | Same as the primary model          | Keep it consistent with the primary model first              | Simple judgment and formatting         | Frequent lightweight steps           | Confirm that the service provider supports the corresponding calling method     |
| Permission Modes    | \[Confirm each time]               | Keep \[Confirm each time] when real files are involved       | Decide whether tools need approval     | File, terminal, and network tasks    | \[Full access] may delete files or access the network                           |
| Heartbeat detection | Enabled, interval 30 minutes       | Can be turned off when there are no ongoing tasks            | Let the Agent check work periodically  | Monitoring, continuous collaboration | This is not a daily report schedule; use scheduled tasks for fixed times        |

## User case: Contract Review Agent

Create a “Contract Review” Agent, choose a verified model as the primary model, require the prompt to output in the format of “risk clause, original location, impact, suggestion,” bind the company policy knowledge base, and keep permissions set to \[Confirm each time]. Create a separate task for each contract and select the corresponding file directory to avoid mixing different clients’ materials in the same context.

<details>

<summary>Why didn’t I see the full MCP and permission settings during creation?</summary>

The creation wizard only keeps the common steps. After creation, open the Agent edit window to continue configuring \[Basics], \[Built-in Tools], \[Knowledge Base], \[MCP], \[Skills], and \[Advanced].

</details>


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.cherryai.com.cn/docs/en-us/advanced-basic/agent-workspace/create-agent.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.