SpaceCat Task Processor

Overview

SpaceCat Task Processor is a Node.js service that processes messages from the AWS SQS queue SPACECAT-TASK-PROCESSOR-JOBS. Based on the type field in each message, it dispatches the message to the appropriate handler for processing various site-related tasks.

Features

• Receives and processes messages from SQS
• Supports multiple task types via modular handlers
• Built-in handlers for audit status, demo URL preparation, disabling imports/audits, generic agent execution, and Slack notifications
• Extensible and easy to add new handlers

Handlers

• opportunity-status-processor: Checks and reports status audits for a site
• disable-import-audit-processor: Disables specified imports and audits for a site
• demo-url-processor: Prepares and shares a demo URL for a site
• agent-executor: Runs registered AI/LLM agents (e.g., the brand-profile agent) asynchronously after onboarding flows
• slack-notify: Sends Slack notifications (text or block messages) from workflows

Setup

1. Clone the repository
2. Install dependencies:

   npm install

3. Configure AWS credentials and environment variables as needed

Usage

• The service is designed to run as a serverless function or background worker.
• It can be invoked in two ways:
  • SQS mode: listens to the SPACECAT-TASK-PROCESSOR-JOBS queue and processes messages automatically (default path for existing workflows).
  • Direct mode: the Lambda entrypoint auto-detects single-message payloads (e.g., from AWS Step Functions) and executes the corresponding handler synchronously. This is used by the new agent workflows to obtain immediate results before triggering follow-up actions.

Development

• To run tests:

  npm test

• To run the optional brand-profile integration test (requires Azure OpenAI env variables):

  npm run test:brand-profile-it

Agent Executor Environment

The agent-executor (and the provided brand-profile agent) rely on the Azure OpenAI credentials consumed by @adobe/spacecat-shared-gpt-client. Ensure the following variables are configured in the Lambda/runner environment (and locally when running the IT test):

Variable	Purpose
AZURE_OPENAI_ENDPOINT	Azure OpenAI endpoint URL
AZURE_OPENAI_KEY	API key for the Azure OpenAI resource
AZURE_API_VERSION	API version used for the chat completions
AZURE_COMPLETION_DEPLOYMENT	Deployment/model name (e.g., gpt-4o)

When invoking the integration test, you can also set BRAND_PROFILE_TEST_BASE_URL to control which site is analyzed and BRAND_PROFILE_IT_FULL=1 to print the complete agent response (otherwise the preview is truncated for readability).

• To lint code:

  npm run lint

Extending

To add a new handler:

1. Create a new folder in src/ for your handler.
2. Export your handler function.
3. Add it to the handler mapping in src/index.js.

---

For more details, see the documentation in src/README.md.

Status

Installation

$ npm install @adobe/spacecat-task-processor

Usage

See the API documentation.

Development

Build

$ npm install

Test

$ npm test

Lint

$ npm run lint

Message Body Formats

Task processor consumes the SPACECAT-TASK-PROCESSOR-JOBS queue, performs the requested task and sends a notification to slack as needed.

Expected message body format in SPACECAT-TASK-PROCESSOR-JOBS is:

{
  "type": "string",
  "siteId": "string"
}