Skip to Content
GlossaryAuto-Configure - Glossary

Auto-Configure

Back to GlossaryConfiguration

An AI-powered feature that generates endpoint request and response mappings by analyzing curl commands, API documentation, or code snippets.

Also known as: auto-configuration, endpoint auto-configure

Overview

Configuring an endpoint in Rhesis requires specifying how to format requests to your API and how to extract fields from its responses. Auto-Configure automates this by using an LLM to analyze existing API documentation or examples and generate the configuration for you.

How It Works

  1. Provide reference material: Paste a curl command, a code snippet, OpenAPI spec excerpt, or a plain-text description of your API.
  2. LLM analysis: The auto-configure service sends the material to an LLM, which analyzes the request and response structure.
  3. Endpoint probing: The service probes your actual endpoint with a generated test request to validate the configuration.
  4. Self-correction: If the probe fails, the service feeds the error back to the LLM, which adjusts and retries—up to a configured number of attempts.
  5. Review and confirm: You review the proposed configuration in a two-step modal before it is saved.

Security Features

API Key Redaction: Auto-Configure automatically detects and redacts API keys and secrets from the reference material before sending it to the LLM. A warning is displayed in the UI when keys are detected.

SSRF Protection: Cloud metadata service addresses (such as 169.254.169.254) are blocked during endpoint probing to prevent Server-Side Request Forgery attacks. Localhost addresses are allowed to support local development.

UI Workflow

The auto-configure modal is integrated into the endpoint creation form and contains two steps:

  1. Input step: Paste your reference material
  2. Review step: Inspect the proposed configuration, edit if needed, and confirm

SDK Usage

python
from rhesis.sdk.entities import Endpoint

# Generate configuration from a curl command
config = Endpoint.auto_configure(
      reference="curl -X POST https://api.example.com/chat -H 'Content-Type: application/json'"
)

Best Practices

  • Review the auto-generated configuration carefully in the review step before saving, especially the response mappings
  • Provide a representative example that includes a real (non-sensitive) request and response for best results
  • Remove or redact API keys from your reference material before pasting—Auto-Configure will warn you if it detects them
  • Test the auto-configured endpoint with a small test set before scaling up to a full test run

Related Terms

TermsAcknowledgments