Agent Integrations

Connect Lamdis to your AI assistants for automated testing.

Supported Integration Methods

Lamdis can test any AI assistant that communicates via HTTP. The primary integration method is:

HTTP Chat API

The most flexible option. Your assistant exposes an HTTP endpoint that accepts messages and returns responses. Works with:

Custom chatbots and agents
RAG systems
Workflow agents
Any AI service with an HTTP API

How Integration Works

Create a Connection in Integrations → Connections
Configure authentication (API key, OAuth, or custom headers)
Create an Assistant in Assistants using the connection
Reference the assistant in your test suites

Request/Response Format

For HTTP Chat integrations, Lamdis sends:


{
  "message": "User's message",
  "transcript": [
    { "role": "user", "content": "Previous message" },
    { "role": "assistant", "content": "Previous response" }
  ],
  "persona": "Optional persona description"
}

Your assistant should respond with:


{
  "reply": "Assistant's response text"
}

Getting Started

Create a connection to your assistant’s endpoint
Set up an assistant configuration
Create a test suite and run your first test

Self-Hosted Assistants

For assistants running locally or in your own infrastructure:

Ensure the endpoint is accessible from Lamdis (or your self-hosted runner)
Use environment-specific connections for dev/staging/prod
Configure any required authentication headers

See Connections for detailed setup instructions.