Understanding agents

Create a standard agent

Agentic studio includes prebuilt agents that automate common marketing tasks. For more specialized needs, you can create a custom agent tailored to your use case.

A standard agent is a flexible, conversational type of custom agent. It determines how to perform tasks based on your instructions and selected tools, context, and skills. Standard agents are best suited for dynamic or less structured scenarios where a fixed workflow is not required.

Use a standard agent when you need a simple, chat-style experience that can adapt to different requests. If your use case requires structured, repeatable, or multi-step execution, consider creating a workflow agent instead.

If you only need to make minor changes to an existing agent, consider duplicating it, adjusting its configuration, or reusing its exported JSON instead of creating a new agent from scratch.

Before you begin

Ensure you have a Builder license. This is required to create custom agents in Agentic studio.
Define what the agent should do and what the user wants to achieve.

To create a standard agent:

In the navigation menu, click Agentic. Then, in the left sidebar, select Agents.
On the Agents page, click Create.
In the Create Agent dialog, select Agent.
On the Create Standard Agent page, provide the following:
- Agent Name - enter a clear, descriptive name.
- Description - add a brief summary of what the agent does.
- Instructions - define the agent’s behavior, personality, and constraints using a system prompt. The system prompt sets how the agent responds, including how it interprets inputs, what rules it follows, and how it structures its output.
Configure capabilities

In the Tools section, select which tools the agent can use when generating a response. The agent determines when to use them. Only tools enabled by an admin in Settings are available for selection.
- Save Artifacts - store generated content as artifacts.
- Context Retrieval - use brand knowledge, guidelines, and other context.
- Web Search - retrieve recent information from the internet.
- Image Generation - create images from text.
In the Agent API section, select which Agent API capabilities the agent can use. Only capabilities enabled by an admin in Settings are available for selection.
- Sites - retrieve sites and site information.
- Pages - create and update pages.
- Content - manage content items.
- Components - manage components and data sources.
- Assets - upload and manage media.
- Personalization - manage personalization variants and conditions.
- Jobs - view and revert background tasks.
- Briefs - list brief types, create and manage briefs.
- Brand kits - list brand kits and retrieve brand kit details.
In the Default Context section, select sources to guide responses. Only sources enabled by an admin in Settings are available for selection.
- Sitecore Documentation MCP - access official Sitecore documentation.
- Marketer MCP - perform SitecoreAI actions for managing sites, pages, content, and more through the Marketer MCP tools.
- Brand Kit - apply brand kit guidelines to responses.
In the Skills section, select skills (custom and system) that the agent can use depending on the task. Only skills enabled by an admin in Settings are available for selection, for example:

Important

If you selected a skill and want to ensure the agent uses it, state this explicitly in the Instructions. For example, add a line Use the Competitive Analysis skill to analyze the user-provided site.
- Campaign planning - plan and structure marketing campaigns by defining objectives, audiences, messaging, channels, and success metrics.
- Competitive analysis - research competitors, compare positioning and messaging, and identify gaps and strategic opportunities.
- CSV - analyze and manipulate CSV data, including filtering, sorting, and summarizing.
- Text - search, extract, and analyze text data, including generating statistics and identifying patterns.
In the Output Format section, choose how responses are returned. Only formats enabled by an admin in Settings are available for selection.
- Text - standard conversational output.
- Structured output (JSON) - formats the response to match a selected schema. You have these options:
  - Select a schema. Available schemas are those created and enabled in Settings.
  - Click Create new schema to add a new JSON schema in Settings.
  - Select Auto-mode to let the agent automatically select the appropriate structure.
Click Create Agent.

Your agent is saved and opens on its run page, where you can provide inputs and run the agent.

Example: Content optimization agent

This is a simple example of a standard agent, including the required inputs and configuration.

Agent configuration

Agent name - Content optimization agent
Description - Analyzes draft marketing content (emails, ads, social posts, landing pages) and suggests improvements for clarity, engagement, and conversion.
Instructions - You are a marketing copy expert focused on improving content performance. Analyze the provided content and suggest specific improvements.

Focus on:
- Clarity and readability
- Strength of headline/hook
- Call-to-action effectiveness
- Tone and audience alignment
- Conciseness and structure
Provide:
- A short summary of issues
- Specific rewrite suggestions
- An improved version of the content
Avoid generic advice. Be specific and actionable. Do not change the core message unless necessary.
Tools - Context retrieval, Save artifacts
Skills - none selected
Output format - Text

Test the agent

After creating the agent, run it using the sample prompt:

Prompt - Analyze this draft content.

You can either paste the content directly or upload a draft file.

Review the output. If the results are not satisfactory, refine the agent by updating its instructions, tools, skills, or context, and test again.

If you have suggestions for improving this article, let us know!