AI Assistant API

DRAFT — Internal Developer Use Only

This API reference is for internal development teams.

Overview

What it is: The AI assistant API provides access to Discovery's AI features: autocomplete for query building, agent sessions for interactive AI conversations, and model management.

Why it matters: Developers building AI-powered features or integrating AI capabilities into other Hydden products need to understand how to create sessions, send queries, and retrieve AI-generated responses.

Endpoints

Method	Path	Description	Auth required
POST	/internal/v1/ai/autocomplete	AI-powered query autocomplete	JWT + API token
GET	/internal/v1/ai/models/:provider	List AI models by provider	JWT + API token
POST	/internal/v1/ai/agent/session	Create a new AI agent session	JWT + API token
GET	/internal/v1/ai/agent/sessions	List all agent sessions	JWT + API token
GET	/internal/v1/ai/agent/sessions/:id	Get a specific agent session	JWT + API token

---

POST /internal/v1/ai/autocomplete

Get AI-powered autocomplete suggestions for search queries and collector configurations.

Request:

POST /internal/v1/ai/autocomplete
Authorization: Bearer <token>
Content-Type: application/json

{
  "input": "find all admin accounts with threat score",
  "context": "global-search"
}

Response (200):

{
  "suggestions": [
    "find all admin accounts with threat score above 80",
    "find all admin accounts with threat score in critical range",
    "find all admin accounts with threat score trending upward"
  ]
}

---

POST /internal/v1/ai/agent/session

Create a new AI agent session for interactive conversations.

Request:

POST /internal/v1/ai/agent/session
Authorization: Bearer <token>
Content-Type: application/json

{
  "context": "universal-collector",
  "initialPrompt": "Help me create a sandboxed Python script for collecting data from a custom REST API"
}

Response (201):

{
  "sessionId": "session-uuid-001",
  "status": "active",
  "createdAt": "2026-02-12T10:00:00Z"
}

---

GET /internal/v1/ai/models/:provider

List available AI models for a given provider.

Path parameters:

Parameter	Type	Description
provider	string	AI provider name

Request:

GET /internal/v1/ai/models/default
Authorization: Bearer <token>

Response (200):

{
  "provider": "default",
  "models": [
    {
      "id": "model-001",
      "name": "Discovery Assistant",
      "capabilities": ["autocomplete", "agent", "search"]
    }
  ]
}