Skip to main content

Assistant API Overview

The Informer Assistant API provides endpoints for creating and managing AI assistants with custom skills, knowledge, and integrations. All routes are prefixed with /api.

Features

  • CRUD Operations - Create, read, update, and delete assistants
  • Skills Management - Add and configure assistant capabilities
  • Secrets - Manage API keys and credentials for integrations
  • Dataset Knowledge - Associate datasets for assistant context
  • References & Files - Attach files and reference documents
  • Image Generation - Generate icons and images
  • Draft Workflow - Edit assistants safely without affecting production
  • Comments - Collaborate on assistant development
  • Sharing - Control access with team-based permissions
  • Tokens - Generate access tokens for assistant APIs

Assistant Types

FieldDescription
CustomUser-created assistants with specific instructions and skills
DispatcherSpecial assistants that route to other assistants
SharedTeam-shared assistants vs personal assistants

Authentication

All Assistant API endpoints require:

  • Valid session or API token
  • AI entitlement verification (ai.verify pre-block)
  • Write permission for modification operations

Common Permission Patterns

OperationPermission Required
Createassistants:create
ReadAI seat (read access implicit)
Updateassistant:write
Deleteassistant:delete
Shareassistant:share

Draft Workflow

Assistants support a draft mode for safe editing:

  1. Create Draft - POST /api/assistants/{id}/_edit
  2. Modify Draft - Changes only affect draft version
  3. Commit Draft - POST /api/assistants/{id}/draft/_commit to publish
  4. Discard Draft - DELETE /api/assistants/{id}/draft to abandon changes

Skills

Skills define what an assistant can do:

  • Query Datasets - Search and analyze data
  • Web Search - Access external information
  • Code Execution - Run Python or other code
  • File Analysis - Process uploaded documents
  • Custom Skills - User-defined capabilities

Dataset Knowledge

Associate datasets to provide domain-specific context:

  • Full Dataset - All records and fields
  • Subset - Filtered records
  • Schema Only - Field definitions without data
  • Sampled - Representative data sample

Error Responses

Standard HTTP status codes:

  • 200 - Success
  • 201 - Created
  • 400 - Bad request (validation error)
  • 403 - Forbidden (insufficient permissions)
  • 404 - Resource not found
  • 500 - Internal server error

Next Steps

Explore the specific endpoint categories: