Data Packs

Data Packs are pre-configured datasets that hydrate Website Environments with realistic initial state. They provide consistent starting conditions for tasks and enable reproducible evaluations.

​

What are Data Packs?

A Data Pack is a collection of data that defines the initial state of an environment. When you load a data pack:

1. The environment resets to a clean state
2. The data pack’s contents populate the database
3. The environment is ready for agent interaction

​

Why Use Data Packs?

---

​

Available Data Packs

Data packs are specific to each Website Environment and provide realistic, pre-configured datasets that simulate different user scenarios.

​

Calendr

​

Cloudfile

​

Shopora

Shopora loads its default product catalog automatically. The default catalog includes:

• 120+ products across Skincare, Makeup, Haircare, Fragrance, and Bath & Body
• 18 popular beauty brands (e.g., Glossier, The Ordinary, Fenty Beauty)
• 35 categories with 2-level hierarchy
• 5 physical stores in major cities (SF, NYC, LA, Chicago, Seattle)
• Pre-populated user with order history, wishlist, and loyalty points
• Product reviews from various users
• Active promo codes for testing discount flows

​

Pandora’s Inbox

---

​

Loading Data Packs

​

Via Web UI

1. Select an environment
2. Choose “Load Task” or reset session
3. Select a data pack from the dropdown
4. Confirm to initialize

​

Via API (Website Environments)

Call the /reset endpoint with the data pack name: See API Reference for details.

---

​

Session Isolation

Data packs work with session isolation:

• Each sessionId gets its own instance
• Multiple sessions can use different data packs
• Sessions don’t share or contaminate data
• Reset a session without affecting others

---

​

Custom Data Packs

---

​

Next Steps