AI Agent

Provides a generic AI agent implementation handling the feedback loop between user requests, tool calls and LLM responses. Compatible with 8.8.0-alpha6 or later.

outbound io.camunda.agenticai:aiagent:1

Required inputs

Key	Label	Type
`provider.anthropic.authentication.apiKey`	Anthropic API key	secret
`provider.bedrock.region`	Region	FEEL expression
`provider.bedrock.authentication.accessKey`	Access key	secret
`provider.bedrock.authentication.secretKey`	Secret key	secret
`provider.openai.authentication.apiKey`	OpenAI API key	secret
`provider.anthropic.model.model`	Model	FEEL expression
`provider.bedrock.model.model`	Model	FEEL expression
`provider.openai.model.model`	Model	FEEL expression
`data.systemPrompt.prompt`	System prompt	FEEL expression
`data.userPrompt.prompt`	User prompt	FEEL expression
`data.agentContext`	Agent context	FEEL expression

Optional inputs

Key	Label	Type
`provider.type`	Provider	string
`provider.bedrock.authentication.type`	Authentication	string
`provider.openai.authentication.organizationId`	Organization ID	FEEL expression
`provider.openai.authentication.projectId`	Project ID	FEEL expression
`provider.openai.endpoint`	Custom API endpoint	FEEL expression
`provider.openai.headers`	Custom headers	FEEL expression
`provider.anthropic.model.parameters.maxTokens`	Maximum tokens	secret
`provider.anthropic.model.parameters.temperature`	Temperature	FEEL expression
`provider.anthropic.model.parameters.topP`	top P	FEEL expression
`provider.anthropic.model.parameters.topK`	top K	FEEL expression
`provider.bedrock.model.parameters.maxTokens`	Maximum tokens	secret
`provider.bedrock.model.parameters.temperature`	Temperature	FEEL expression
`provider.bedrock.model.parameters.topP`	top P	FEEL expression
`provider.openai.model.parameters.maxCompletionTokens`	Maximum completion tokens	secret
`provider.openai.model.parameters.temperature`	Temperature	FEEL expression
`provider.openai.model.parameters.topP`	top P	FEEL expression
`data.systemPrompt.parameters`	System prompt parameters	FEEL expression
`data.userPrompt.parameters`	User prompt parameters	FEEL expression
`data.userPrompt.documents`	Documents	FEEL expression
`data.tools.containerElementId`	Ad-hoc sub-process ID	FEEL expression
`data.tools.toolCallResults`	Tool call results	FEEL expression
`data.memory.storage.type`	Memory storage type	string
`data.memory.storage.timeToLive`	Document TTL	FEEL expression
`data.memory.storage.customProperties`	Custom document properties	FEEL expression
`data.memory.contextWindowSize`	Context window size	string
`data.limits.maxModelCalls`	Maximum model calls	string
`data.response.format.type`	Response format	string
`data.response.format.parseJson`	Parse text as JSON	string
`data.response.format.schema`	Response JSON schema	FEEL expression
`data.response.format.schemaName`	Response JSON schema name	FEEL expression
`data.response.includeAssistantMessage`	Include assistant message	string
`resultVariable`	Result variable	string
`resultExpression`	Result expression	FEEL expression
`errorExpression`	Error expression	FEEL expression
`retryCount`	Retries	FEEL expression
`retryBackoff`	Retry backoff	string

Apply this connector with @bpmnkit/connectors

import { applyConnectorTemplate } from "@bpmnkit/connectors" const result = applyConnectorTemplate("io.camunda.connectors.agenticai.aiagent.v1", { "provider.anthropic.authentication.apiKey": "<your anthropic api key>", "provider.bedrock.region": "= ...", "provider.bedrock.authentication.accessKey": "<your access key>", "provider.bedrock.authentication.secretKey": "<your secret key>", "provider.openai.authentication.apiKey": "<your openai api key>", "provider.anthropic.model.model": "= claude-3-5-sonnet-20240620", "provider.bedrock.model.model": "= anthropic.claude-3-5-sonnet-20240620-v1:0", "provider.openai.model.model": "= gpt-4o", "data.systemPrompt.prompt": "= You are **TaskAgent**, a helpful, generic chat agent that can handle a wide variety of customer requests using your own domain knowledge **and** any tools explicitly provided to you at runtime. If tools are provided, you should prefer them instead of guessing an answer. You can call the same tool multiple times by providing different input values. Don't guess any tools which were not explicitely configured. If no tool matches the request, try to generate an answer. If you're not able to find a good answer, return with a message stating why you're not able to. Wrap minimal, inspectable reasoning in *exactly* this XML template: <thinking> <context>…briefly state the customer’s need and current state…</context> <reflection>…list candidate tools, justify which you will call next and why…</reflection> </thinking> Reveal **no** additional private reasoning outside these tags. ", "data.userPrompt.prompt": "= ...", "data.agentContext": "= =agent.context", "provider.type": "anthropic", "provider.bedrock.authentication.type": "defaultCredentialsChain", })