API Reference

Below are our AIDR endpoints that are currently in private beta. Reach out if you are interested.

Analyze Input:

  • Endpoint: /v1/analyze/input
  • Summary: Analyzes LLM input and runs the selected detectors(e.g. adversarial input detection). An example input would be a webpage your LLM is summarizing.

Analyze Output

  • Endpoint: /v1/analyze/output
  • Summary: Analyzes LLM output and runs the selected detectors(e.g. suspicious URL detection). An example output would be a summary of a webpage.

Analyze Actions

  • Endpoint: /v1/analyze/action
  • Summary: Analyzes LLM actions(function call & params) and runs the selected detectors(e.g. anomaly detection). An example action would be a writing to a database or sending an email.

Detectors

  • Adversarial Input
  • Anomaly Detection
  • PII
  • Suspicious URLs
  • Data Exfiltration Vectors
  • Secrets
  • Invisible Unicode
  • And many more!