Agentic Data Plane

Create an Agent

Create a new AI agent declaratively through Redpanda Agentic Data Plane. No Python or JavaScript code required. This guide walks through the managed create canvas section by section, then covers how to connect to and test the running agent.

After reading this page, you will be able to:

  • Create a managed Agentic Data Plane agent through the create canvas

  • Configure an LLM provider, model, and tools for an agent

  • Connect to and test a running agent

Prerequisites

Open the create canvas

  1. Open Agents in the sidebar.

  2. Click Create agent.

  3. Choose how the agent runs. Click Redpanda manages it, so Redpanda deploys, runs, and observes the agent for you. (To run an agent you host yourself, see Set Up a Self-Managed Agent.)

    The runtime choice in the create-agent flow

Agentic Data Plane opens the New managed agent canvas: a single page with sections for Identity, Model, Instructions, and Tools, plus a collapsible Advanced area for subagents and tags. Everything on the canvas is editable later from the agent’s Overview tab. Each field is marked Required or Optional; the Create agent button stays disabled until you provide the required fields, with helper text explaining what’s still missing.

Identity

In the Identity section, name the agent and optionally describe it.

  1. Enter a Name. Required. Shown in the agent registry list and the detail header (for example, Customer Support Agent). Agentic Data Plane derives the agent’s resource ID from this name when you create the agent; you don’t enter the ID directly. The derived ID is a lowercase slug that starts with a letter, ends with a letter or digit, and can contain letters, digits, and hyphens (up to 63 characters). The ID is used in URLs and CLI commands and can’t be changed after creation.

    Click Suggest a name to generate a placeholder you can edit.

  2. Optionally, add a Description. An internal note shown on the agent’s detail page. This is not the system prompt; write that under Instructions.

Model

In the Model section, choose the large language model (LLM) provider and model the agent uses to interpret user intent and decide which tools to invoke. Provider and model are selected together in one picker: pick a provider to see the models it serves, then pick a model. The picker lists every LLM provider configured in Agentic Data Plane (for example, support-bedrock-v2) and the models each one exposes (for example, us.anthropic.claude-haiku-4-5-20251001-v1:0). Calls to the model are proxied through the AI Gateway.

To add a new provider, see Configure an LLM provider. For model selection guidance, see Model selection guide.

For external model specifications and pricing:

Max iterations is not on the create canvas. It’s the upper bound on agent reasoning loops per request, and you set it after creation from the agent’s Overview tab, in the Model section. Leave it unset (or 0) to use the runtime default of 100. Each iteration consumes tokens and adds latency, so tune it to task complexity: simple queries (10-20), balanced workflows (20-40), complex analysis (40-100).

For an AWS Bedrock provider, Agentic Data Plane validates the model when you create or update the agent. A model ID that isn’t a valid Bedrock model for the provider’s region is rejected at save time, with an error that points at the offending model field, instead of being accepted and then failing later when the agent deploys. A common mistake is the bare foundation-model ID: Claude 4.6 and later require the full inference-profile ID (for example, us.anthropic.claude-opus-4-7 rather than claude-opus-4-7). If you type a model the picker doesn’t list, enter the full inference-profile ID. The same check applies to each subagent’s model override. See AWS Bedrock: Inference profiles and IAM.

Instructions

In the Instructions section, write the System prompt that defines the agent’s role, responsibilities, constraints, and output format. The system prompt is optional but strongly recommended; when set, it’s sent on every invocation as the first message. The field is a Markdown editor with a Write tab and a Preview tab. On the Write tab, type the prompt directly and use the formatting toolbar for headings, bold, italic, links, lists, quotes, and code blocks, or select the upload icon to import a Markdown file. The Preview tab renders the formatted result. To edit in a larger surface, click Expand editor, and to start from a template, select Templates.

A typical prompt defines:

  • Agent role and responsibilities

  • Available tools (matches the servers you select in the Tools section)

  • Safety rules and constraints

  • Expected output format

Example structure:

You are an [agent role].

Responsibilities:
- [Task 1]
- [Task 2]

Available tools:
- [tool_name]: [description]

Never:
- [Constraint 1]
- [Constraint 2]

Response format:
- [Format guideline]

For full guidance on writing effective prompts, see System prompt best practices.

Tools

In the Tools section, grant the agent access to the Model Context Protocol (MCP) servers that expose tools to it. This section is optional, and you can change the list at any time after creation.

  1. In the Search MCP servers box, find each server you want the agent to call and select its checkbox. To create a new server instead of attaching an existing one, click Add MCP server, which opens the server-creation flow in a new tab.

  2. The agent discovers each selected server’s tools at runtime; you don’t pre-select individual tools. The agent calls them based on the LLM’s interpretation of the system prompt and the user request. Keep the system prompt aligned with the servers you select so the model knows what’s available.

Apply the principle of least privilege. Connect only the servers the agent needs.

An agent can reference at most 32 MCP servers in total, counting the servers on the root agent plus every subagent’s servers. The limit counts references, not distinct servers, so a server referenced by both the root agent and a subagent counts twice. If the total exceeds 32, you can’t save the agent.

For registering MCP servers, see Create an MCP server.

Advanced: subagents and tags

Expand the Advanced area to add subagents and tags. Both are optional.

Subagents

Subagents are internal specialists within the same agent. Each subagent has its own name, delegation hint, system prompt, and MCP tools. By default, a subagent inherits the parent agent’s model and LLM provider, but you can override both per subagent. All subagents share the parent agent’s provider credentials, gateway endpoint, and execution settings. The root agent orchestrates and delegates work to the appropriate subagent based on the request.

To add a specialist, click Add subagent (or Add a subagent) and configure it in the panel that opens:

  • Name: Required. Must start with a lowercase letter, contain only lowercase letters, digits, and hyphens, and end with a letter or digit (for example, researcher). The name must be unique among the agent’s subagents.

  • When should the parent use it?: Required. The delegation hint the parent reads to decide when to route a task to this subagent. Be specific about inputs, outputs, and use cases.

  • Model: Optional. Defaults to the parent agent’s model. Pick a different provider and model in the picker to override it for this subagent; use Reset to inherited to go back to the parent’s model.

  • Instructions: Required. Domain-specific system prompt for what the subagent does.

  • Tools: Optional. Select the MCP servers this subagent can access.

Click Add subagent to save it, or Cancel to discard. To keep the agent flat, leave the list empty.

For multi-agent design patterns, see Agent architecture patterns.

Tags

Add Tags as key/value pairs to organize and filter agents in the registry and to break down Cost & Usage by team, environment, and more.

Create the agent

  1. Review the canvas. The Create agent button is disabled until you enter a name and pick a model; the helper text next to it tells you which is still missing.

  2. Click Create agent.

  3. Wait for the agent to reach the Running state.

When the agent is running, open its detail page. A managed agent has these tabs:

  • Overview: The agent’s full configuration as an editable form. Change the description and tags (in the Details section; the name is read-only there), the model, the system prompt (in the Behavior section), tools, and subagents, then save.

  • Connect: The agent’s A2A endpoint URL, authentication guidance, and ready-to-paste code examples for calling the agent. See Connect to the agent.

  • Triggers: Schedules and event sources that invoke the agent. See Trigger Agents from External Channels.

  • Inspector: An interactive test surface. See Test your agent.

  • Cost & Usage: Spend, tokens, and latency for the agent.

  • Transcripts: A record of every conversation against the agent.

Connect to the agent

A running managed agent exposes an A2A endpoint. Open the agent’s Connect tab to get everything you need to call it programmatically or integrate it with external systems.

  • A2A endpoint: The URL where the agent receives messages. Copy it from the Connect tab. (The URL is provisioned once the agent starts.)

  • Authentication: The endpoint accepts Redpanda Cloud bearer tokens. Create a service account, exchange its client credentials for an access token, and send the token as an Authorization: Bearer header on every request. Use Manage service accounts on the Connect tab to open Organization IAM > Service accounts. For how service accounts and role bindings work, see Service account authorization.

  • Code examples: Select a language to get a ready-to-paste sample that calls the agent with an official A2A client SDK (Go, Node.js, Python, and Java), or a cURL walkthrough.

For more about calling agents and integrating them with other systems, see Choose an Integration Pattern.

Update a managed agent

Change a managed agent’s configuration at any time after you create it, either in Agentic Data Plane or with the rpk ai agent update command. You can edit settings such as the model, system prompt, tools, and subagents. The agent ID is fixed and can’t change.

Updates roll out without interrupting the agent. Redpanda starts the new version and waits for it to become ready before it stops the old one, so in-flight and new requests keep reaching a running instance throughout the change.

Configure A2A discovery metadata (optional)

A2A discovery metadata lets external systems find and invoke the agent through capability-based discovery.

For agent card design guidance, see Create an Agent Card.

Configure identity and provider fields for the agent card:

  • Icon URL: A publicly accessible image URL (recommended: 256x256 PNG or SVG).

  • Documentation URL: Link to comprehensive agent documentation.

  • Organization: Your organization or team name.

  • URL: Website or contact URL.

Add Skills to describe what the agent can do for capability-based discovery. For each skill:

  • Skill ID (required): Lowercase letters, digits, and hyphens (for example, fraud-analysis).

  • Skill name (required): Human-readable name displayed in agent directories.

  • Description (required): What the skill does and when to use it. Be specific about inputs, outputs, and use cases.

  • Tags (optional): Categorization terms (for example, fraud, security, finance).

  • Examples (optional): Sample queries demonstrating how to invoke this skill.

The metadata is published at https://<your-agent-url>/.well-known/agent-card.json.

For more about agent cards and how they’re consumed, see Agent cards.

Test your agent

Two tabs on the agent detail page support testing and observability:

  • Inspector: Send test prompts interactively and inspect each step of the agent’s reasoning, tool calls, and responses. Use Inspector during development and tuning. You can also override the model or max iterations for a single test run here.

  • Transcripts: Records every conversation against the agent (prompt, tool calls, tool outputs, final response). Use Transcripts to audit production behavior.

To test the agent in Inspector:

  1. Open the agent’s Inspector tab.

  2. Enter a test prompt.

  3. Verify the agent:

    • Selects appropriate tools

    • Follows the system prompt’s constraints

    • Returns the expected output format

  4. Iterate on the system prompt, MCP tool selection, or model choice as needed.

For detailed observability strategies, see Monitor Agent Activity.

Example configurations

Simple query agent

  • Model: GPT-5 Mini (fast, cost-effective)

  • Tools: Single MCP server exposing a get_orders tool

  • Use case: Customer order lookups

Complex analytics agent

  • Model: Claude Sonnet 4.5 (balanced)

  • Tools: Multiple servers for data query, aggregation, and formatting

  • Use case: Multi-step data analysis

Multi-agent orchestrator

  • Model: Claude Opus 4.5 (advanced reasoning)

  • Tools: Tools the root agent delegates to subagents

  • Subagents: order-agent, inventory-agent, customer-agent

  • Use case: E-commerce operations

Next steps