An agent is a configured conversation runner. It holds a system prompt, a voice, an LLM model, and a set of tools / playbooks / FAQs it can draw on. End-users talk to agents — never directly to the underlying model.Documentation Index
Fetch the complete documentation index at: https://docs.kataven.ai/llms.txt
Use this file to discover all available pages before exploring further.
What’s on an agent
| Field | Notes |
|---|---|
name, description, category | Display only. |
status | draft / active / paused / archived. Only active agents accept calls. |
system_prompt | The full instructional prompt fed to the LLM. |
greeting_message | First thing the agent says. |
llm_model | gpt-4o, claude-sonnet-4-6, etc. |
voice_provider + voice_id | Cartesia / ElevenLabs / your provider here. |
greeting_mode | interruptible (caller can talk over the greeting) or locked. |
ringing_enabled | Whether to play a synthesized ring before pickup. |
background_audio_enabled + background_audio_volume | Office-ambience track behind the agent’s voice. |
noise_suppression_enabled | DTLN inbound noise suppression. |
auth_mode | public (anyone can talk) or authenticated (requires a signed widget JWT). |
playbooks, faqs | Attached knowledge — see Playbooks & FAQs. |
Lifecycle
Onlyactive agents accept inbound calls and originate outbound calls.
Operations
| Operation | Endpoint | SDK |
|---|---|---|
| List | GET /api/agents | client.agents.list() |
| Get | GET /api/agents/{id} | client.agents.get(id) |
| Create | POST /api/agents | client.agents.create(...) |
| Update | PATCH /api/agents/{id} | client.agents.update(id, ...) |
| Delete | DELETE /api/agents/{id} | client.agents.delete(id) |
| Attach playbook | POST /api/agents/{id}/playbooks | client.agents.attach_playbook(id, pb_id) |
| Detach playbook | DELETE /api/agents/{id}/playbooks/{pbId} | client.agents.detach_playbook(id, pb_id) |
| Attach FAQ | POST /api/agents/{id}/faqs | client.agents.attach_faq(id, faq_id) |
| Configure tool | POST /api/agents/{id}/tools | client.agents.upsert_tool(id, tool_name, config, enabled) |