Document updated on Feb 11, 2026

Amazon Web Services Bedrock integration

The Bedrock interface allows KrakenD to use Bedrock’s Converse API without writing custom integration code, enabling intelligent automation, content generation, or any LLM-powered use case within your existing API infrastructure.

This component abstracts you from the Bedrock API, allowing the consumer to focus solely on the prompt. For each request to an endpoint, KrakenD will create the Bedrock request with all the necessary elements in its API and return a unified response, so if you use other vendors, you have a consistent use of LLM models.

In other words, the user sends content, like “tell me a joke!”, and KrakenD builds the API payload needed to talk to Bedrock.

This Bedrock interface configures a backend within KrakenD that transparently forwards REST requests to Bedrock’s API endpoints. It manages authentication, versioning, and payload formatting using its custom templating system. This way, you can easily call Bedrock models without writing custom integration code.

A simple configuration looks like this:

{
  "endpoint": "/bedrock",
  "method": "POST",
  "backend": [
    {
      "host": [
        "https://bedrock-runtime.eu-west-1.amazonaws.com"
      ],
      "url_pattern": "/model/eu.meta.llama3-2-1b-instruct-v1:0/converse",
      "method": "POST",
      "extra_config": {
        "ai/llm": {
          "bedrock": {
            "v1": {
              "credentials": "bedrock-api-key-xxx",
              "debug": true,
              "variables": {
                "max_tokens": 1500
              }
            }
          }
        }
      }
    }
  ]
}

To interact with the LLM, the user can send in the request:

• instructions (optional): If you want to add a system prompt
• contents: The content you want to send to the template

Like this:

$curl -XPOST --json '{"instructions": "Act as a 1000 dollar consultant", "contents": "Tell me a consultant joke"}' http://localhost:8080/bedrock

Configuration of Bedrock

The Bedrock configuration requires you to add the ai/llm namespace under your backend’ extra_config’, with the Bedrock vendor.

Customizing the payload sent and received from Bedrock

As with all KrakenD LLM interfaces, you can completely replace the request and response to have a custom interaction with the LLM. While the default template should allow you to accomplish any day-to-day job, you might need to extend it using your own template.

You may override the input and output Go templates by specifying:

• input_template: Path to a custom template controlling how the request data is formatted before sending to Bedrock.
• output_template: Path to a custom template to transform and extract the desired pieces from Bedrock’s response.

See below how to change this interaction.

Default input_template for Bedrock v1

When you don’t set any input_template, KrakenD will create the JSON payload sent to Bedrock using the following template:

{
	"inferenceConfig": {
		{{ $max_tokens := .variables.max_tokens }}{{ if ge $max_tokens 0 }}"maxTokens": {{ $max_tokens }},{{ end }}
		{{ $temperature := .variables.temperature }}{{ if ge $temperature 0.0 }}"temperature": {{ $temperature }},{{ end }}
		{{ $top_p := .variables.top_p }}{{ if ge $top_p 0.0 }}"topP": {{ $top_p }},{{ end }}
		{{ $top_k := .variables.top_k }}{{ if ge $top_k 0 }}"topK": {{ $top_k }},{{ end }}
		"stopSequences": {{ .variables.stop_sequences | toJson }}
	},
	{{- if hasKey .req_body "instructions" }}
	"system": [
		{
			"text": {{ .req_body.instructions | toJson }}
		}
	],
	{{ end }}
	"messages": [
		{
			"content": [
				{
					"text": {{ .req_body.contents | toJson }}
				}
			],
			"role": "user"
		}
	]
}

Remember, you can access your own variables declared in the configuration using {{ .variables.xxx }}.

Default output_template for Bedrock v1

When you don’t declare an output_template, the response from the AI is transformed to have the following format:

{
	"ai_gateway_response":
	[
		{{ if gt (len .resp_body.output.message.content) 0 }}
		{
			"contents": [
			{{ range $ci, $content := .resp_body.output.message.content }}
				{{ if $ci }},{{ end }}
				{{ $content.text | toJson }}
			{{ end }}
			]
		}
		{{ end }}
	],
	"usage": "{{ .resp_body.usage.totalTokens }}"
}