Guardrly

Guardrly observes every API request made by an AI agent that interacts with Shopify or Meta Ads, creating a real‑time audit log without requiring code changes. It records the HTTP method, endpoint, status and timing of each call, and translates raw requests into plain‑English operation labels such as “Deleted product Vintage Chair.” Sensitive information is automatically scrubbed before any data leaves the host machine.

When a request triggers a platform risk—like a rapid Ads API call that could cause a ban, or a high‑risk deletion in Shopify—Guardrly sends immediate alerts via email. The service also captures the agent’s decision reasoning for high‑risk actions, providing evidence that can be used for appeals or troubleshooting.

Installation consists of a single command that configures the monitoring layer for compatible AI tools, including Claude Code, Cursor, and other MCP‑compatible agents. The resulting audit trail and decision logs give operators visibility into the agent’s behavior, helping prevent unexpected account bans or unintended changes.