Official
E-commerce Solutions
Builds secure, admin-facing AI agents within Medusa projects to automate store operations and data analysis.
This skill provides a comprehensive framework for developing internal AI agents integrated directly into the Medusa admin dashboard. It guides developers through the entire stack—from defining shared data models for sessions and messages to implementing secure MedusaExec capabilities for data operations and building streaming React UI components. It ensures all agents remain strictly admin-facing, maintaining high security through authenticated routes and sandboxed execution, making it ideal for creating tools like product auditors, cohort analysis assistants, or automated store operation managers.
Key Features
01NDJSON streaming support for real-time agent responses in the Medusa Admin
02Shared infrastructure for AgentSession and AgentMessage models with agent_type filtering
03Admin dashboard UI extensions with chat interfaces and session management
04Secure MedusaExec tool for sandboxed TypeScript execution on store data
05Per-agent configuration for custom system prompts and specialized capabilities
06197 GitHub stars
Use Cases
01Building a store operations assistant for bulk product auditing and metadata updates
02Creating a merchant-facing cohort analysis agent to identify customer buying patterns
03Developing internal support tooling for staff to query complex order and return data