Skip to main content
POST
Create Agent Configuration

Body

application/json

Request model for creating agent configurations (POST /agent-configurations).

Used only by the create endpoint. The update endpoint (PUT /{id}) accepts AgentUpdate instead. model_name and agent_name are required for create (validated in endpoint). hub_config, api_key_id, agent_policy_files are ignored for backward compat.

id
string | null
hub
string
default:custom
model_name
string | null
agent_name
string | null
agent_url
string | null
hub_config
Hub Config · object | null
status
string
default:active
api_key_id
string | null
api_key
string | null
agent_system_prompt
string | null
agent_policy_files
string[] | null
rag_params
Rag Params · object | null
function_route_params
Function Route Params · object | null
mcp_config
MCPConfig · object | null

MCP (Model Context Protocol) proxy configuration.

Response

Successful Response

Agent domain model.

id
string<uuid>
required
agent_name
string
required
model_name
string
required
team_id
string<uuid>
required
created_by
string<uuid>
required
created_at
integer
required
updated_at
integer
required
agent_url
string | null
hub
string
default:custom
api_key
string | null
agent_system_prompt
string | null
rate_limit_requests_per_minute
integer
default:100

Rate limit for API requests per minute

Required range: x >= 1
status
enum<string>
default:active

Agent status values.

Available options:
draft,
creating,
active,
under_review,
deprecated,
archived,
pending,
rejected
access_level
enum<string>
default:black_box

Access level for agent analysis.

Black Box: URL only - behavioral testing only Grey Box: Agent configuration - structure/capabilities analysis White Box: Source code access - full code + sandbox runtime analysis

Available options:
black_box,
grey_box,
white_box
agent_params
AgentParams · object | null

Model parameters for agent configuration.

mcp_config
MCPConfig · object | null

MCP (Model Context Protocol) proxy configuration.

delegated_agents
DelegatedAgent · object[]
white_box_config
WhiteBoxConfig · object | null

Source code access configuration for white box analysis.

agent_metadata
AgentMetadata · object | null

Agent metadata for display and organization.

purpose
string | null
protocol
enum<string>
default:chat_completions

Communication protocol for agent endpoints.

Available options:
chat_completions,
a2a
test_score
string | null
Pattern: ^(?!^[-+.]*$)[+-]?0*\d*\.?\d*$
last_tested_by
string | null
evaluated_harnesses
string[] | null
last_evaluated_at
integer | null
protection_status
enum<string>
default:unprotected

Dome protection lifecycle state.

Available options:
unprotected,
configured,
domed
trust_stage
enum<string>
default:registered

Trust maturity lifecycle stage for an agent.

Available options:
registered,
tested,
hardened,
trusted,
optimized,
adapted
trusted_at
integer | null
ttt_days
number | null
dome_config_id
string<uuid> | null
dome_manually_configured
boolean
default:false
import_metadata
ImportMetadata · object

Metadata about how an agent was imported.

capabilities
AgentCapabilities · object

Agent capability flags.

skills
(AgentSkill · object | object)[]

An agent skill definition (for A2A protocol agents).

latest_evaluation
AgentLatestEvaluation · object | null

Latest completed evaluation scores for an agent.

Sideloaded on GET /agent-configurations when include_scores=true. Contains only domain-level score data — no UI or presentation concerns.

Last modified on April 21, 2026