Submit feedback for a task

cURL

curl --request POST \
  --url https://api.{environment}.corti.app/v2/agentic/contexts/{contextId}/tasks/{taskId}/feedback \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --header 'Tenant-Name: <api-key>' \
  --data '
{
  "rating": {
    "scale": "binary",
    "value": 1
  }
}
'

import { CortiClient, CortiEnvironment } from "@corti/sdk";

const client = new CortiClient({
    environment: CortiEnvironment.Eu,
    auth: {
        clientId: "YOUR_CLIENT_ID",
        clientSecret: "YOUR_CLIENT_SECRET"
    },
    tenantName: "YOUR_TENANT_NAME"
});
await client.agentic.contexts.tasks.feedback.create("ctx.0192f4c8-3d6b-7c4f-a02b-4d9e7f3c8b51", "task.0192f4c8-4e7c-7d50-b13c-5eaf8a4d9c62", {
    rating: {
        scale: "binary",
        value: 0
    },
    labels: ["unsupportedClaim"],
    reason: "The response stated that the patient had diabetes, but this was not present in the available data.",
    target: {
        messageId: "msg.0192f4c8-5f8d-7e61-924d-6fb09b5ead73"
    },
    metadata: {
        collectionMethod: "caseReview",
        clientReference: "case-review-728193",
        actor: {
            externalId: "clinician_4182"
        }
    }
});

using Corti.Agentic.Contexts.Tasks;
using Corti;

var client = new CortiClient(
    "TENANT_NAME",
    CortiClientEnvironment.Eu,
    new CortiClientAuth.ClientCredentials("client_id", "client_secret")
);
await client.Agentic.Contexts.Tasks.Feedback.CreateAsync(
    "ctx.0192f4c8-3d6b-7c4f-a02b-4d9e7f3c8b51",
    "task.0192f4c8-4e7c-7d50-b13c-5eaf8a4d9c62",
    new AgenticFeedbackCreateRequest
    {
        Rating = new AgenticFeedbackRating { Scale = AgenticFeedbackRatingScale.Binary, Value = 0 },
        Labels = new List<AgenticFeedbackLabel>() { AgenticFeedbackLabel.UnsupportedClaim },
        Reason =
            "The response stated that the patient had diabetes, but this was not present in the available data.",
        Target = new AgenticFeedbackTarget
        {
            MessageId = "msg.0192f4c8-5f8d-7e61-924d-6fb09b5ead73",
        },
        Metadata = new AgenticFeedbackMetadata
        {
            CollectionMethod = "caseReview",
            ClientReference = "case-review-728193",
            Actor = new AgenticFeedbackActor { ExternalId = "clinician_4182" },
        },
    }
);

import requests

url = "https://api.{environment}.corti.app/v2/agentic/contexts/{contextId}/tasks/{taskId}/feedback"

payload = { "rating": {
        "scale": "binary",
        "value": 1
    } }
headers = {
    "Authorization": "Bearer <token>",
    "Tenant-Name": "<api-key>",
    "Content-Type": "application/json"
}

response = requests.post(url, json=payload, headers=headers)

print(response.text)

<?php

$curl = curl_init();

curl_setopt_array($curl, [
  CURLOPT_URL => "https://api.{environment}.corti.app/v2/agentic/contexts/{contextId}/tasks/{taskId}/feedback",
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_ENCODING => "",
  CURLOPT_MAXREDIRS => 10,
  CURLOPT_TIMEOUT => 30,
  CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  CURLOPT_CUSTOMREQUEST => "POST",
  CURLOPT_POSTFIELDS => json_encode([
    'rating' => [
        'scale' => 'binary',
        'value' => 1
    ]
  ]),
  CURLOPT_HTTPHEADER => [
    "Authorization: Bearer <token>",
    "Content-Type: application/json",
    "Tenant-Name: <api-key>"
  ],
]);

$response = curl_exec($curl);
$err = curl_error($curl);

curl_close($curl);

if ($err) {
  echo "cURL Error #:" . $err;
} else {
  echo $response;
}

package main

import (
	"fmt"
	"strings"
	"net/http"
	"io"
)

func main() {

	url := "https://api.{environment}.corti.app/v2/agentic/contexts/{contextId}/tasks/{taskId}/feedback"

	payload := strings.NewReader("{\n  \"rating\": {\n    \"scale\": \"binary\",\n    \"value\": 1\n  }\n}")

	req, _ := http.NewRequest("POST", url, payload)

	req.Header.Add("Authorization", "Bearer <token>")
	req.Header.Add("Tenant-Name", "<api-key>")
	req.Header.Add("Content-Type", "application/json")

	res, _ := http.DefaultClient.Do(req)

	defer res.Body.Close()
	body, _ := io.ReadAll(res.Body)

	fmt.Println(string(body))

}

HttpResponse<String> response = Unirest.post("https://api.{environment}.corti.app/v2/agentic/contexts/{contextId}/tasks/{taskId}/feedback")
  .header("Authorization", "Bearer <token>")
  .header("Tenant-Name", "<api-key>")
  .header("Content-Type", "application/json")
  .body("{\n  \"rating\": {\n    \"scale\": \"binary\",\n    \"value\": 1\n  }\n}")
  .asString();

{
  "id": "fb.0192f4c8-9a3b-7e2f-9c4d-5a6b7c8d9e0f",
  "taskId": "task.0192f4c8-4e7c-7d50-b13c-5eaf8a4d9c62",
  "rating": {
    "scale": "binary",
    "value": 1
  },
  "normalizedScore": 1,
  "labels": [],
  "createdAt": "2026-05-19T12:00:00Z"
}

{
  "error": {
    "code": "ASSIGNMENT_CONFLICT",
    "message": "could not complete assignment",
    "details": {
      "assignment_id": "asg_99",
      "expert_id": "exp_42"
    }
  }
}

{
  "error": {
    "code": "ASSIGNMENT_CONFLICT",
    "message": "could not complete assignment",
    "details": {
      "assignment_id": "asg_99",
      "expert_id": "exp_42"
    }
  }
}

{
  "error": {
    "code": "ASSIGNMENT_CONFLICT",
    "message": "could not complete assignment",
    "details": {
      "assignment_id": "asg_99",
      "expert_id": "exp_42"
    }
  }
}

{
  "error": {
    "code": "ASSIGNMENT_CONFLICT",
    "message": "could not complete assignment",
    "details": {
      "assignment_id": "asg_99",
      "expert_id": "exp_42"
    }
  }
}

POST

agentic

contexts

{contextId}

tasks

{taskId}

feedback

cURL

curl --request POST \
  --url https://api.{environment}.corti.app/v2/agentic/contexts/{contextId}/tasks/{taskId}/feedback \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --header 'Tenant-Name: <api-key>' \
  --data '
{
  "rating": {
    "scale": "binary",
    "value": 1
  }
}
'

import { CortiClient, CortiEnvironment } from "@corti/sdk";

const client = new CortiClient({
    environment: CortiEnvironment.Eu,
    auth: {
        clientId: "YOUR_CLIENT_ID",
        clientSecret: "YOUR_CLIENT_SECRET"
    },
    tenantName: "YOUR_TENANT_NAME"
});
await client.agentic.contexts.tasks.feedback.create("ctx.0192f4c8-3d6b-7c4f-a02b-4d9e7f3c8b51", "task.0192f4c8-4e7c-7d50-b13c-5eaf8a4d9c62", {
    rating: {
        scale: "binary",
        value: 0
    },
    labels: ["unsupportedClaim"],
    reason: "The response stated that the patient had diabetes, but this was not present in the available data.",
    target: {
        messageId: "msg.0192f4c8-5f8d-7e61-924d-6fb09b5ead73"
    },
    metadata: {
        collectionMethod: "caseReview",
        clientReference: "case-review-728193",
        actor: {
            externalId: "clinician_4182"
        }
    }
});

using Corti.Agentic.Contexts.Tasks;
using Corti;

var client = new CortiClient(
    "TENANT_NAME",
    CortiClientEnvironment.Eu,
    new CortiClientAuth.ClientCredentials("client_id", "client_secret")
);
await client.Agentic.Contexts.Tasks.Feedback.CreateAsync(
    "ctx.0192f4c8-3d6b-7c4f-a02b-4d9e7f3c8b51",
    "task.0192f4c8-4e7c-7d50-b13c-5eaf8a4d9c62",
    new AgenticFeedbackCreateRequest
    {
        Rating = new AgenticFeedbackRating { Scale = AgenticFeedbackRatingScale.Binary, Value = 0 },
        Labels = new List<AgenticFeedbackLabel>() { AgenticFeedbackLabel.UnsupportedClaim },
        Reason =
            "The response stated that the patient had diabetes, but this was not present in the available data.",
        Target = new AgenticFeedbackTarget
        {
            MessageId = "msg.0192f4c8-5f8d-7e61-924d-6fb09b5ead73",
        },
        Metadata = new AgenticFeedbackMetadata
        {
            CollectionMethod = "caseReview",
            ClientReference = "case-review-728193",
            Actor = new AgenticFeedbackActor { ExternalId = "clinician_4182" },
        },
    }
);

import requests

url = "https://api.{environment}.corti.app/v2/agentic/contexts/{contextId}/tasks/{taskId}/feedback"

payload = { "rating": {
        "scale": "binary",
        "value": 1
    } }
headers = {
    "Authorization": "Bearer <token>",
    "Tenant-Name": "<api-key>",
    "Content-Type": "application/json"
}

response = requests.post(url, json=payload, headers=headers)

print(response.text)

<?php

$curl = curl_init();

curl_setopt_array($curl, [
  CURLOPT_URL => "https://api.{environment}.corti.app/v2/agentic/contexts/{contextId}/tasks/{taskId}/feedback",
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_ENCODING => "",
  CURLOPT_MAXREDIRS => 10,
  CURLOPT_TIMEOUT => 30,
  CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  CURLOPT_CUSTOMREQUEST => "POST",
  CURLOPT_POSTFIELDS => json_encode([
    'rating' => [
        'scale' => 'binary',
        'value' => 1
    ]
  ]),
  CURLOPT_HTTPHEADER => [
    "Authorization: Bearer <token>",
    "Content-Type: application/json",
    "Tenant-Name: <api-key>"
  ],
]);

$response = curl_exec($curl);
$err = curl_error($curl);

curl_close($curl);

if ($err) {
  echo "cURL Error #:" . $err;
} else {
  echo $response;
}

package main

import (
	"fmt"
	"strings"
	"net/http"
	"io"
)

func main() {

	url := "https://api.{environment}.corti.app/v2/agentic/contexts/{contextId}/tasks/{taskId}/feedback"

	payload := strings.NewReader("{\n  \"rating\": {\n    \"scale\": \"binary\",\n    \"value\": 1\n  }\n}")

	req, _ := http.NewRequest("POST", url, payload)

	req.Header.Add("Authorization", "Bearer <token>")
	req.Header.Add("Tenant-Name", "<api-key>")
	req.Header.Add("Content-Type", "application/json")

	res, _ := http.DefaultClient.Do(req)

	defer res.Body.Close()
	body, _ := io.ReadAll(res.Body)

	fmt.Println(string(body))

}

HttpResponse<String> response = Unirest.post("https://api.{environment}.corti.app/v2/agentic/contexts/{contextId}/tasks/{taskId}/feedback")
  .header("Authorization", "Bearer <token>")
  .header("Tenant-Name", "<api-key>")
  .header("Content-Type", "application/json")
  .body("{\n  \"rating\": {\n    \"scale\": \"binary\",\n    \"value\": 1\n  }\n}")
  .asString();

{
  "id": "fb.0192f4c8-9a3b-7e2f-9c4d-5a6b7c8d9e0f",
  "taskId": "task.0192f4c8-4e7c-7d50-b13c-5eaf8a4d9c62",
  "rating": {
    "scale": "binary",
    "value": 1
  },
  "normalizedScore": 1,
  "labels": [],
  "createdAt": "2026-05-19T12:00:00Z"
}

{
  "error": {
    "code": "ASSIGNMENT_CONFLICT",
    "message": "could not complete assignment",
    "details": {
      "assignment_id": "asg_99",
      "expert_id": "exp_42"
    }
  }
}

{
  "error": {
    "code": "ASSIGNMENT_CONFLICT",
    "message": "could not complete assignment",
    "details": {
      "assignment_id": "asg_99",
      "expert_id": "exp_42"
    }
  }
}

{
  "error": {
    "code": "ASSIGNMENT_CONFLICT",
    "message": "could not complete assignment",
    "details": {
      "assignment_id": "asg_99",
      "expert_id": "exp_42"
    }
  }
}

{
  "error": {
    "code": "ASSIGNMENT_CONFLICT",
    "message": "could not complete assignment",
    "details": {
      "assignment_id": "asg_99",
      "expert_id": "exp_42"
    }
  }
}

Authorizations

Authorization

string

header

required

OAuth 2.0 / OIDC bearer token.

Tenant-Name

string

header

required

The tenant the request operates within.

Path Parameters

contextId

string<context-id>

required

Context identifier (prefixed UUIDv7). Context identifier. Accepts ctx.<uuidv7> or a bare UUIDv7 on input; always returned prefixed.

Pattern: ^(ctx\.)?[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$

Example:

"ctx.0192f4c8-3d6b-7c4f-a02b-4d9e7f3c8b51"

taskId

string<task-id>

required

Task identifier (prefixed UUIDv7). Task identifier. Accepts task.<uuidv7> or a bare UUIDv7 on input; always returned prefixed.

Pattern: ^(task\.)?[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$

Example:

"task.0192f4c8-4e7c-7d50-b13c-5eaf8a4d9c62"

Body

application/json

The feedback to submit.

Request body for submitting task feedback.

rating

object

required

The original rating supplied by the customer.

Show child attributes

Example:

{ "scale": "binary", "value": 1 }

labels

enum<string>[]

Structured observations about the result. Defaults to an empty array. Positive and negative labels may be combined. Duplicate labels are rejected. A maximum of five labels may be submitted.

Maximum array length: 5

Structured observation about the result. Positive and negative labels share one taxonomy so customers can represent mixed feedback.

correct — factually and contextually correct (positive).
complete — included the important expected information (positive).
helpful — materially helped the user complete their task (positive).
wellPresented — clear, readable, appropriately structured (positive).
efficient — reached a useful result without unnecessary content (positive).
incorrect — one or more claims, conclusions, or actions were wrong (negative).
missingInformation — important or expected information was omitted (negative).
irrelevant — included content that did not address the request (negative).
misunderstoodRequest — the system responded to the wrong intent (negative).
unsupportedClaim — a claim not supported by available information (negative).
unsafeOrInappropriate — unsafe, disallowed, or unsuitable (negative).
poorlyPresented — difficult to read or unsuitably structured (negative).
tooVerbose — substantially more detail than useful (negative).
other — another observation described in reason (both).

Available options:

correct,

complete,

helpful,

wellPresented,

efficient,

incorrect,

missingInformation,

irrelevant,

misunderstoodRequest,

unsupportedClaim,

unsafeOrInappropriate,

poorlyPresented,

tooVerbose,

other

reason

string

The user's explanation of the rating or labels. Required when labels contains other.

Maximum string length: 1000

Examples:

"The response stated the patient had diabetes"

"but this was not present in the available data."

target

object

Identifies the specific user-visible response being evaluated. If omitted, the feedback applies to the task as a whole.

Show child attributes

metadata

object

Customer-provided provenance and correlation information.

Show child attributes

Response

Feedback created.

A stored feedback resource.

string<feedback-id>

required

Feedback identifier. Accepts fb.<uuidv7> or a bare UUIDv7 on input; always returned prefixed.

Pattern: ^(fb\.)?[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$

Example:

"fb.0192f4c8-7e2a-7b3c-9d4e-5f6a7b8c9d01"

taskId

string<task-id>

required

Task identifier. Accepts task.<uuidv7> or a bare UUIDv7 on input; always returned prefixed.

Pattern: ^(task\.)?[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$

Example:

"task.0192f4c8-4e7c-7d50-b13c-5eaf8a4d9c62"

rating

object

required

The original rating supplied by the customer.

Show child attributes

Example:

{ "scale": "binary", "value": 1 }

normalizedScore

number<double>

required

Corti-derived internal score between 0 and 1. The original scale and value are always retained alongside this score.

Required range: 0 <= x <= 1

Example:

1

labels

enum<string>[]

required

Structured observations about the result.

Structured observation about the result. Positive and negative labels share one taxonomy so customers can represent mixed feedback.

correct — factually and contextually correct (positive).
complete — included the important expected information (positive).
helpful — materially helped the user complete their task (positive).
wellPresented — clear, readable, appropriately structured (positive).
efficient — reached a useful result without unnecessary content (positive).
incorrect — one or more claims, conclusions, or actions were wrong (negative).
missingInformation — important or expected information was omitted (negative).
irrelevant — included content that did not address the request (negative).
misunderstoodRequest — the system responded to the wrong intent (negative).
unsupportedClaim — a claim not supported by available information (negative).
unsafeOrInappropriate — unsafe, disallowed, or unsuitable (negative).
poorlyPresented — difficult to read or unsuitably structured (negative).
tooVerbose — substantially more detail than useful (negative).
other — another observation described in reason (both).

Available options:

correct,

complete,

helpful,

wellPresented,

efficient,

incorrect,

missingInformation,

irrelevant,

misunderstoodRequest,

unsupportedClaim,

unsafeOrInappropriate,

poorlyPresented,

tooVerbose,

other

Example:

["correct", "helpful"]

createdAt

string<date-time>

required

read-only

When the feedback was created.

Example:

"2026-05-19T12:00:00Z"

reason

string

Free-text explanation of the rating or labels.

Maximum string length: 1000

Examples:

"The response stated the patient had diabetes"

"but this was not present in the available data."

target

object

Identifies the specific user-visible response being evaluated. If omitted, the feedback applies to the task as a whole.

Show child attributes

metadata

object

Customer-provided provenance and correlation information.

Show child attributes

Web Socket APIs

REST APIs

Agentic APIs v2 (beta)

Agentic APIs v1 (deprecated)

Admin APIs

Corti Models APIs

Errors

Authorizations

Path Parameters

Body

Response