Skip to main content
POST
/
api
/
v1
/
qa-agents
Create a QA agent
curl --request POST \
  --url https://{subdomain}.mihu.ai/api/v1/qa-agents \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data @- <<EOF
{
  "name": "Support QA Agent",
  "description": "Evaluates support conversations for empathy and compliance",
  "evaluates_agent_type": "human",
  "language": "English",
  "is_default": true,
  "skills": [
    {
      "name": "Greeting",
      "weight": 15,
      "description": "Professional opening and introduction"
    },
    {
      "name": "Empathy",
      "weight": 20,
      "description": "Understanding and acknowledging customer concerns",
      "positive_items": [
        {
          "name": "Used the customer's name",
          "points": 5
        }
      ],
      "penalty_items": [
        {
          "name": "Interrupted the customer",
          "penalty_points": -5
        }
      ]
    },
    {
      "name": "Compliance",
      "weight": 25,
      "description": "Adherence to regulatory and company policies"
    },
    {
      "name": "Resolution",
      "weight": 25,
      "description": "Effective problem solving and issue resolution"
    },
    {
      "name": "Sales",
      "weight": 15,
      "description": "Upselling and cross-selling opportunities"
    }
  ]
}
EOF
{
  "success": true,
  "message": "QA agent created successfully",
  "data": {
    "uuid": "7e1c9b3a-5d2f-4c8e-b6a1-0f9d8c7b6a5e",
    "name": "Support QA Agent",
    "description": "Evaluates support conversations for empathy and compliance",
    "evaluates_agent_type": "human",
    "is_active": true,
    "is_default": true,
    "evaluation_model": "advanced",
    "strictness_level": 70,
    "confidence_threshold": 85,
    "auto_fail_threshold": 3,
    "language": "English",
    "custom_instructions": "Focus on product knowledge when evaluating technical support calls.",
    "automatic_scoring": true,
    "sentiment_analysis": true,
    "compliance_checking": true,
    "skills": [
      {
        "uuid": "9c2f5a44-1b7e-4f4a-9a6c-3d2f8e7b1c0a",
        "name": "Empathy",
        "description": "Understanding and acknowledging customer concerns",
        "weight": 20,
        "max_score": 10,
        "ai_evaluation_prompt": "Evaluate whether the agent acknowledged the customer's feelings.",
        "positive_items": [
          {
            "name": "Used the customer's name",
            "points": 5
          }
        ],
        "penalty_items": [
          {
            "name": "Interrupted the customer",
            "penalty_points": -5,
            "is_auto_fail": false
          }
        ],
        "order": 2,
        "is_custom": false
      }
    ],
    "total_weight": 100,
    "weights_balanced": true,
    "created_at": "2026-06-04T10:15:30+00:00",
    "updated_at": "2026-06-04T10:15:30+00:00"
  }
}

Authorizations

​
Authorization
string
header
required

Use a Bearer token to access these API endpoints. Example: "Bearer {your-token}"

Body

application/json

Create/update payload. On update only the fields you send are changed.

​
name
string
required
Maximum string length: 255
Example:

"Support QA Agent"

​
description
string | null
Example:

"Evaluates support conversations for empathy and compliance"

​
evaluates_agent_type
enum<string>
default:both

Who this QA agent evaluates: human agents, AI agents, or both.

Available options:
human,
ai,
both
Example:

"human"

​
is_active
boolean
default:true
Example:

true

​
is_default
boolean
default:false

Send true to make this the default evaluator. To move the default away from an agent, set another one as default.

Example:

false

​
evaluation_model
string
default:advanced
Maximum string length: 50
Example:

"advanced"

​
strictness_level
integer
default:70

How strictly conversations are scored. Higher = harsher.

Required range: 0 <= x <= 100
Example:

70

​
confidence_threshold
integer
default:85

Minimum AI confidence (%) for a score to be applied.

Required range: 0 <= x <= 100
Example:

85

​
auto_fail_threshold
integer
default:3

Number of critical violations after which an evaluation fails automatically.

Required range: 0 <= x <= 10
Example:

3

​
language
string
default:English

Language for evaluation reports and feedback.

Maximum string length: 50
Example:

"English"

​
custom_instructions
string | null

Extra guidance applied to every evaluation.

Example:

"Focus on product knowledge when evaluating technical support calls."

​
automatic_scoring
boolean
default:true
Example:

true

​
sentiment_analysis
boolean
default:true
Example:

true

​
compliance_checking
boolean
default:true
Example:

true

​
skills
object[]

The full scorecard. Sending it replaces the existing skills; weights must total 100%. Leave it out on create to start with the standard 5-skill scorecard.

Minimum array length: 1

Response

QA agent created successfully

​
success
boolean
Example:

true

​
message
string
Example:

"QA agent created successfully"

​
data
object

A QA agent with its full scorecard.