Tagging Agent — DevOps skill for Claude Code

Support Agent AWS

Automated customer support ticket enrichment system built on AWS serverless infrastructure. When a new Help Scout conversation arrives, the system automatically looks up the customer in Acme, enriches the ticket with organization data, marketplace integrations, tags, and research notes.

Architecture

Help Scout Webhook → API Gateway → webhook-handler → Step Functions
                                                        ├→ acme-lookup
                                                        ├→ integration-fetcher
                                                        ├→ property-updater
                                                        ├→ conversation-tagger
                                                        └→ note-creator

Slack (#ai-training) → API Gateway → slack-orchestrator → DynamoDB

Lambda Functions

Step Functions Pipeline

Function	Purpose
webhook-handler	Receives Help Scout webhooks, validates payload, filters spam/staff emails, starts Step Functions
acme-lookup	Searches Acme for customer organizations (DESIGN/ENDUSER) by email
integration-fetcher	Fetches Etsy, Shopify, WooCommerce integrations for designer orgs
property-updater	Updates Help Scout customer properties (account type, org ID, shop URLs, affiliate status)
conversation-tagger	Tags conversations using keyword rules, dynamic DynamoDB rules, and Claude AI classification
note-creator	Creates research notes with customer links, order history, translations, and dynamic capabilities

Standalone

Function	Purpose
slack-orchestrator	Listens to #ai-training Slack channel for messages starting with #, uses Claude to extract tagging rules and research capabilities, stores them in DynamoDB

Shared Libraries

• @support-agent-aws/shared - Types (ProcessingContext, Organization, Integration), secrets manageme