🤖Agent Builder

Edit Agent from the Side Panel

1. Open Agent Builder from the Side Panel.

2. Select an existing agent using the dropdown menu at the top

In the edit view, update any of the following:

• Name: The display name of your agent.

• Description: A short summary of what the agent does.

• Instructions: The system-level instructions that define your agent’s behavior, tone, task focus, etc.

• Model selection: Choose the AI model your agent will run on

• Capabilities:

  • 🧮 Run Code: Allows the agent to execute Python code using a code interpreter (if enabled).

  • 🖼 Upload File Context (OCR): Let the agent extracting data from scanned documents or images using optical character recognition (OCR).

  • 🔍 File Search: Enable agent to search through uploaded files and respond with relevant content.

• Artifacts & Custom Prompt Mode

  • 🧩 Enable Artifacts: When enabled, the agent can output structured artifacts like code blocks, tasks, or rich results.

  • ⚙️ Include shadcn/ui instructions: Internal flag often used for UI-driven instructions – toggle if your app needs UI guidance. (Optional for most users)

  • ✍️ Custom Prompt Mode: Lets you customize how prompts are formatted or sent to the model. Keep enabled for agents with tailored instruct-style workflows.

• Add Tools

1. Click “Add Tools” under the Tools + Actions section.

2. Browse the available tools.

3. Click the green “Add +” button to enable a tool.

• Add Action

Connect your agent to external APIs for more advanced, dynamic behavior via OpenAPI schemas.

1. Click “Add Actions” under Tools + Actions.

2. Fill in the following fields:

   • Authentication: Choose method (optional, based on API type).

   • Schema: Paste your OpenAPI schema here.

   • Privacy Policy URL: Link to your API's privacy page (required for trust & security).

3. Click Create to add the API action.

• Advanced Setting

  • 🧩 Max Agent Steps: Controlled by the system to limit how many steps the agent can take in a single task.

  • 🔗 Agent Chain (Mixture-of-Agents): Let multiple agents work together in sequence to solve complex tasks.