What is Guardrly?
Guardrly is an MCP-based API interceptor that monitors every API call your AI Agent makes to external platforms like Shopify and Meta Ads.
The Problem
AI Agents like Claude Code and Cursor can execute hundreds of API calls autonomously. Without visibility into these operations, you risk:
- Unexpected data modifications (deleted products, cancelled orders)
- Platform bans from rate limit violations
- No audit trail when something goes wrong
How Guardrly Works
Guardrly installs as an MCP Server alongside your AI Agent. It sits between your Agent and external APIs, silently observing every request:
- Your Agent calls make_http_request
- Guardrly intercepts the call before it leaves your machine
- PII is automatically scrubbed (API keys, emails, tokens)
- The operation is logged to your local SQLite cache
- Logs are shipped to the cloud dashboard
- Risk analysis runs against 100+ semantic rules
- Alerts fire instantly for dangerous operations
Key Principles
Non-invasive: Zero code changes to your Agent or workflows.
Privacy-first: Sensitive data is scrubbed before leaving your machine.
Real-time: Alerts fire within seconds of a dangerous operation.