Skip to main content
POST
cURL

Authorizations

Authorization
string
header
required

Token-based authentication with required prefix "Token"

Body

name
string
required
Required string length: 1 - 255
model
string
required
Required string length: 1 - 255
adapters
string<uuid>[]
template
string<uuid> | null
tags
string[]
Required string length: 1 - 255
timezone
string
Required string length: 1 - 255
system_prompt
string
sleep_until
string<date-time> | null
email_alias
string | null

Custom email alias for the agent's AgentMail inbox (e.g. 'support' becomes support@domain)

Maximum string length: 255
status
enum<string>
  • running - Running
  • sleeping - Sleeping
  • archived - Archived
Available options:
running,
sleeping,
archived
run_started_at
string<date-time> | null
last_wake_reason
string
Maximum string length: 64
last_wake_interrupted_sleep_until
string<date-time> | null
max_turns
integer
Required range: -2147483648 <= x <= 2147483647
max_sleep_hours
integer | null

Optional cap on how long the agent may sleep, in hours. When set, each harness run is started with --max-utc-sleep = now + this many hours, and sleep requests past that moment are clamped.

Required range: 0 <= x <= 2147483647
harness_git_ref
string | null

Optional git ref (branch/tag/commit) for harness runs. Null = latest main at dispatch time.

Maximum string length: 255
reasoning_effort
  • none - None
  • minimal - Minimal
  • low - Low
  • medium - Medium
  • high - High
  • xhigh - Extra high
Available options:
none,
minimal,
low,
medium,
high,
xhigh
debounce_window_seconds
integer | null

Per-agent override for the turn-boundary debounce window (seconds). Leave blank to inherit from the template.

Required range: 0 <= x <= 2147483647
vapi_voice_id
string

Unused: VAPI calls now use the gpt-realtime-2 model, which only supports a fixed set of OpenAI voices and doesn't support per-agent voice overrides. This field has no effect on call behavior.

Maximum string length: 255
purpose
enum<string>
  • production - Production
  • eval - Eval
  • dev - Dev
Available options:
production,
eval,
dev
parent
string<uuid> | null

Response

201 - application/json
id
string<uuid>
required
read-only
organization
string<uuid>
required
read-only
parent_name
string
required
read-only
template_name
string | null
required
read-only
name
string
required
Maximum string length: 255
model
string
required
Maximum string length: 255
created_at
string<date-time>
required
read-only
adapters
string<uuid>[]
template
string<uuid> | null
tags
string[]
Maximum string length: 255
timezone
string
Maximum string length: 255
system_prompt
string
sleep_until
string<date-time> | null
email_alias
string | null

Custom email alias for the agent's AgentMail inbox (e.g. 'support' becomes support@domain)

Maximum string length: 255
status
enum<string>
  • running - Running
  • sleeping - Sleeping
  • archived - Archived
Available options:
running,
sleeping,
archived
run_started_at
string<date-time> | null
last_wake_reason
string
Maximum string length: 64
last_wake_interrupted_sleep_until
string<date-time> | null
max_turns
integer
Required range: -2147483648 <= x <= 2147483647
max_sleep_hours
integer | null

Optional cap on how long the agent may sleep, in hours. When set, each harness run is started with --max-utc-sleep = now + this many hours, and sleep requests past that moment are clamped.

Required range: 0 <= x <= 2147483647
harness_git_ref
string | null

Optional git ref (branch/tag/commit) for harness runs. Null = latest main at dispatch time.

Maximum string length: 255
reasoning_effort
  • none - None
  • minimal - Minimal
  • low - Low
  • medium - Medium
  • high - High
  • xhigh - Extra high
Available options:
none,
minimal,
low,
medium,
high,
xhigh
debounce_window_seconds
integer | null

Per-agent override for the turn-boundary debounce window (seconds). Leave blank to inherit from the template.

Required range: 0 <= x <= 2147483647
vapi_voice_id
string

Unused: VAPI calls now use the gpt-realtime-2 model, which only supports a fixed set of OpenAI voices and doesn't support per-agent voice overrides. This field has no effect on call behavior.

Maximum string length: 255
purpose
enum<string>
  • production - Production
  • eval - Eval
  • dev - Dev
Available options:
production,
eval,
dev
parent
string<uuid> | null