For AI agents: a documentation index is available at the root level at /llms.txt. Append /llms.txt to any URL for a page-level index, or .md for the markdown version of any page.
Creates a new agent using AI generation. Accepts either a guided questionnaire (array of Q&A pairs) or a free-text description. The AI generates the agent's prompt and configuration. A security check is run on the generated prompt before the agent is persisted. Deducts credits on success.
**Rules:**
- Either `questions` (non-empty array) or `description` must be provided — not both.
- `emotiveToggle`, `voiceId`, and `voiceModel` must either all be present or all be absent.
- When `emotiveToggle=true`, `voiceModel` must be `GPT_REALTIME`.
- When `emotiveToggle=false`, `voiceModel` cannot be `GPT_REALTIME`.
Authentication
AuthorizationBearer
API key from the console ApiKey collection, sent as Bearer token. Also accepts session cookies for browser-based auth.
Request
This endpoint expects an object.
namestringOptional<=50 characters
Agent name. Auto-generated if omitted.
descriptionstringOptional<=15000 characters
Free-text description. Required if questions not given.
questionslist of objectsOptional
Guided Q&A pairs. Required if description not given.
typeenumOptional
Workflow type. Only single_prompt is supported.
Allowed values:
emotiveTogglebooleanOptional
Enable emotive (GPT Realtime) mode. Must be given with voiceId and voiceModel.
voiceIdstringOptional
Voice ID. Must be given together with emotiveToggle and voiceModel.
voiceModelstringOptional
Voice model. Must be given together with emotiveToggle and voiceId.
knowledgeBaseIdstringOptional
Knowledge base to attach (must belong to the organization).
Response
Agent created — returns the new agent’s ObjectId
successboolean
datastring
MongoDB ObjectId of the created agent
Errors
400
Bad Request Error
401
Unauthorized Error
403
Forbidden Error
500
Internal Server Error
503
Service Unavailable Error
Creates a new agent using AI generation. Accepts either a guided questionnaire (array of Q&A pairs) or a free-text description. The AI generates the agent’s prompt and configuration. A security check is run on the generated prompt before the agent is persisted. Deducts credits on success.
Rules:
Either questions (non-empty array) or description must be provided — not both.
emotiveToggle, voiceId, and voiceModel must either all be present or all be absent.
When emotiveToggle=true, voiceModel must be GPT_REALTIME.
When emotiveToggle=false, voiceModel cannot be GPT_REALTIME.