DynamoFL API (3.24.3)

Download OpenAPI specification:Download

Users

Get current user

Get the currently authenticated user

Responses

Create user

Create a new user

Request Body schema: application/json
required

email required	string Email address of the user
password	string Password for the user (optional)
orgLevelRoleNames required	Array of strings Array of organization level role names. Valid roles: org:dynamoeval:admin, org:dynamoeval:developer, org:dynamoeval:member, org:dynamoguard:admin, org:dynamoguard:developer, org:dynamoguard:member, org:dynamoguard:extensionUser, org:dynamoguard:extensionAdmin, org:iam:admin, org:iam:editor, org:iam:viewer, org:platform:admin

Responses

Request samples

Payload

Content type

application/json

{"email": "user@example.com",
"password": "SecurePassword123!",
"orgLevelRoleNames": ["org:dynamoeval:admin",
"org:dynamoguard:developer",
"org:iam:viewer"
]
}

Response samples

201

Content type

application/json

{"email": "user@example.com",
"temporaryPassword": "TempPassword123!",
"keycloakId": "123e4567-e89b-12d3-a456-426614174000",
"dynamoAIId": "507f1f77bcf86cd799439011"
}

Get current user with Keycloak info

Get the currently authenticated user with Keycloak information

Responses

Reset password for non-onboarded user

Reset the password for a user who has not completed onboarding

Request Body schema: application/json
required

email required	string Email address of the user
password	string New password for the user (optional)

Responses

Request samples

Payload

Content type

application/json

{"email": "user@example.com",
"password": "NewSecurePassword123!"
}

Response samples

201

Content type

application/json

{"email": "user@example.com",
"temporaryPassword": "TempPassword123!"
}

Reset password for onboarded user

Reset the password for a user who has completed onboarding

Request Body schema: application/json
required

email required	string Email address of the user
currentPassword required	string Current password of the user
newPassword required	string New password for the user

Responses

Request samples

Payload

Content type

application/json

{"email": "user@example.com",
"currentPassword": "CurrentPassword123!",
"newPassword": "NewSecurePassword123!"
}

Get all users

Get all users with optional filtering and organization level roles

query Parameters

includeOrgLevelRoles	boolean Include organization level roles in the response
search	string Search term to filter users

Responses

Response samples

200

Content type

application/json

{"total": 100,
"users": [{"keycloakId": "123e4567-e89b-12d3-a456-426614174000",
"dynamoAIId": "507f1f77bcf86cd799439011",
"email": "user@example.com",
"firstName": "John",
"lastName": "Doe",
"orgLevelRoles": ["Admin"
],
"enabled": true
}
]
}

Update organization level role

Update the organization level roles for a user

path Parameters

id

required

string

User ID

Request Body schema: application/json
required

orgLevelRoles

required

Array of strings

Array of organization level role names. Valid roles: org:dynamoeval:admin, org:dynamoeval:developer, org:dynamoeval:member, org:dynamoguard:admin, org:dynamoguard:developer, org:dynamoguard:member, org:dynamoguard:extensionUser, org:dynamoguard:extensionAdmin, org:iam:admin, org:iam:editor, org:iam:viewer, org:platform:admin

Responses

Request samples

Payload

Content type

application/json

{"orgLevelRoles": ["org:dynamoeval:admin",
"org:dynamoguard:developer",
"org:iam:viewer"
]
}

Delete user

Delete a user by ID

path Parameters

id

required

string

User ID to delete

Responses

Get current user organization level scopes

Get the organization level scopes for the currently authenticated user

Responses

Get current user terms consent status

Get the terms and consent status for the currently authenticated user

Responses

Response samples

200

Content type

application/json

{ }

Update current user terms consent

Update the terms and consent status for the currently authenticated user

Request Body schema: application/json
required

Array of terms consent updates

Array

object (UpdateUserTermsConsentDto)

Responses

Request samples

Payload

Content type

application/json

[{ }
]

Create API token

Create an API token for the currently authenticated user

Responses

Response samples

201

Content type

application/json

{"apiKey": "string"
}

TermsConsentController_getAllTermsConsents

Responses

ModerationController_generatePolicyBehaviors

Request Body schema: application/json
required

object (GenerateBehaviorsDto)

Responses

Request samples

Payload

Content type

application/json

{ }

PolicyController_getPolicyResourceLevelRoles

Responses

PolicyController_addUseCase

Request Body schema: application/json
required

object (AddUseCaseDto)

Responses

Request samples

Payload

Content type

application/json

{ }

PolicyController_getUseCases

Responses

PolicyController_updateUsecase

path Parameters

id

required

string

Request Body schema: application/json
required

object (AddUseCaseDto)

Responses

Request samples

Payload

Content type

application/json

{ }

PolicyController_getUsecase

path Parameters

id

required

string

Responses

PolicyController_deleteUsecase

path Parameters

id

required

string

Responses

PolicyController_getAnyUsecase

path Parameters

id

required

string

Responses

InClusterModelController_upsertModel

path Parameters

deploymentId

required

string

Request Body schema: application/json
required

deploymentStatus required	string Enum: "COMPLETED" "FAILED" "IN_PROGRESS" "DEPLOYMENT_DOES_NOT_EXIST" "UNKNOWN" New status of the model deployment
reason	string Reason of the model deployment

Responses

Request samples

Payload

Content type

application/json

{"deploymentStatus": "COMPLETED",
"reason": "DeploymentReady"
}

InClusterModelController_getAllModels

Responses

InClusterModelController_getModel

path Parameters

deploymentId

required

string

Responses

InClusterModelController_deleteModel

path Parameters

deploymentId

required

string

Responses

DynamoGuard Analyze

Apply policies on messages

header Parameters

X-Num-Tokens	string Number of tokens analyzed by policies
Warning	string Warning displayed about tokens being not analyzed due to length

Request Body schema: application/json
required

required	Array of objects (AnalyzeMessage) OpenAI style messages with optional RAG context to analyze
textType required	string Enum: "MODEL_INPUT" "MODEL_RESPONSE" Whether last message is a prompt or response
policyIds required	Array of strings List of policies to apply
inheritPolicies	boolean Default: false Whether to inherit policies from the modelId passed
modelId	string The model id of the LLM to associate inference with
clientId	string The client id of the sender
metadata	object Metadata to be attached to the request

Responses

Request samples

Payload

Content type

application/json

{"messages": [{"role": "user",
"content": "string",
"ragContext": "string"
}
],
"textType": "MODEL_INPUT",
"policyIds": ["string"
],
"inheritPolicies": false,
"modelId": "string",
"clientId": "string",
"metadata": { }
}

Response samples

200

Content type

application/json

{"text": "string",
"textType": "MODEL_INPUT",
"finalAction": "BLOCK",
"appliedPolicies": [{"policy": {"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
]
},
"outputs": { },
"action": "string"
}
],
"error": "string"
}

ExtensionController_analyzeText

Request Body schema: application/json
required

required	Array of objects (AnalyzeMessage) OpenAI style messages with optional RAG context to analyze
textType required	string Enum: "MODEL_INPUT" "MODEL_RESPONSE" Whether last message is a prompt or response
inheritPolicies	boolean Default: false Whether to inherit policies from the modelId passed
modelId	string The model id of the LLM to associate inference with
clientId	string The client id of the sender
metadata	object Metadata to be attached to the request
domain required	string Domain of the chatbot the user is interacting with

Responses

Request samples

Payload

Content type

application/json

{"messages": [{"role": "user",
"content": "string",
"ragContext": "string"
}
],
"textType": "MODEL_INPUT",
"inheritPolicies": false,
"modelId": "string",
"clientId": "string",
"metadata": { },
"domain": "string"
}

Response samples

200

Content type

application/json

{"text": "string",
"textType": "MODEL_INPUT",
"finalAction": "BLOCK",
"appliedPolicies": [{"policy": {"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
]
},
"outputs": { },
"action": "string"
}
],
"error": "string"
}

DynamoGuard Logs

Get Logs

Retrieve logs from chat and analyze interactions. Narrow search down by using the query filters.

path Parameters

modelId

required

string

query Parameters

startTime	number The start time (in UTC seconds) to filter by
endTime	number The end time (in UTC seconds) to filter by
pageNum	number Default: 1 The page number
perPage	number Default: 50 The number of results per page
sortDir	string Default: "desc" Enum: "asc" "desc" The direction to sort results by.
policyId	string The policy id to filter by
policyType	string Enum: "PII" "TOXICITY" "ALIGNMENT" "HALLUCINATION" "RAG_HALLUCINATION" "INPUT_RELEVANCE" "RESPONSE_RELEVANCE" "OOB_HALLUCINATION" "KEYWORD" The policy type to filter by
actions	Array of strings List of actions to filter by
textType	string Enum: "MODEL_INPUT" "MODEL_RESPONSE" The text type to filter by
action	string Enum: "BLOCK" "WARN" "REDACT" "SANITIZE" "NONE" The policy action to filter by
clientId	string The client id to filter by
userId	string The user id to filter by
policyTypes	Array of strings List of policy types to filter by
metadata	string Example: metadata={origin: {platform: "chatgpt.com", browser: "Chrome"}, version: "3.16"} User-defined log metadata to filter by. Should be a json string

Responses

Response samples

200

Content type

application/json

{"logs": [{"timestamp": "string",
"analyses": [{"text": "string",
"textType": "MODEL_INPUT",
"appliedPolicies": [{"outputs": { },
"violated": true,
"feedback": { },
"promptId": "string",
"policy": {"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"methodParams": { },
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
]
},
"action": "BLOCK"
}
]
}
],
"action": "BLOCK",
"metadata": { }
}
],
"totalCount": 0,
"warnings": [{"name": "Token Length Exceeded Warning",
"message": "string"
}
]
}

Get Count of Logs

Retrieve number of log entries from chat and analyze interactions. Narrow search down by using the query filters.

path Parameters

modelId

required

string

query Parameters

startTime	number The start time (in UTC seconds) to filter by
endTime	number The end time (in UTC seconds) to filter by
policyId	string The policy id to filter by
policyType	string Enum: "PII" "TOXICITY" "ALIGNMENT" "HALLUCINATION" "RAG_HALLUCINATION" "INPUT_RELEVANCE" "RESPONSE_RELEVANCE" "OOB_HALLUCINATION" "KEYWORD" The policy type to filter by
actions	Array of strings List of actions to filter by
textType	string Enum: "MODEL_INPUT" "MODEL_RESPONSE" The text type to filter by
action	string Enum: "BLOCK" "WARN" "REDACT" "SANITIZE" "NONE" The policy action to filter by
clientId	string The client id to filter by
userId	string The user id to filter by
policyTypes	Array of strings List of policy types to filter by
metadata	string Example: metadata={origin: {platform: "chatgpt.com", browser: "Chrome"}, version: "3.16"} User-defined log metadata to filter by. Should be a json string

Responses

Response samples

200

Content type

application/json

0

0

Get unique metadata fields

Return the complete set of unique metadata fields observed for this model across its lifetime.

path Parameters

modelId

required

string

The AI System (model) id

Responses

Response samples

200

Content type

application/json

{"fields": ["sessionId",
"traceId",
"order.id"
]
}

Get Logs by User Email and Time Range

Retrieve logs for a specific user within a given time range. The maximum allowed time range is 1 month (30 Days).

path Parameters

userEmail

required

string

query Parameters

startTime required	number
endTime required	number
sortDir required	string

Responses

Response samples

200

Content type

application/json

{"presignedUrl": "string",
"logCount": 0
}

DynamoGuard Stats

Get stats

Retrieve dashboard metrics. Narrow search down by using the query filters. At least one of modelId, clientId, or userId must be specified.

path Parameters

modelId

required

string

query Parameters

startTime	number The start time to filter by
endTime	number The end time to filter by
numDataPoints	number Default: 2 Number of data points to discretize time period. Must be greater than 1.
clientId	string The client id to filter by

Responses

Response samples

200

Content type

application/json

{"overview": {"MODEL_INPUT": {"total": [0
],
"flagged": [0
]
},
"MODEL_RESPONSE": {"total": [0
],
"flagged": [0
]
}
},
"privacy": {"MODEL_INPUT": {"entity_types": {"LOCATION": [0,
37
]
},
"total": [0
]
},
"MODEL_RESPONSE": {"entity_types": {"LOCATION": [0,
37
]
},
"total": [0
]
}
},
"keyword": {"MODEL_INPUT": {"entity_types": {"lawyer": [0,
37
]
},
"total": [0
]
},
"MODEL_RESPONSE": {"entity_types": {"lawyer": [0,
37
]
},
"total": [0
]
}
},
"toxicity": {"MODEL_INPUT": {"total": [0
],
"violation_rate": {"value": 0,
"data": [0
]
}
},
"MODEL_RESPONSE": {"total": [0
],
"violation_rate": {"value": 0,
"data": [0
]
}
}
},
"alignment": {"MODEL_INPUT": {"total": [0
],
"policies": {"property1": {"name": "string",
"num_violations": [0
],
"violation_rate": {"value": 0,
"data": [0
]
}
},
"property2": {"name": "string",
"num_violations": [0
],
"violation_rate": {"value": 0,
"data": [0
]
}
}
}
},
"MODEL_RESPONSE": {"total": [0
],
"policies": {"property1": {"name": "string",
"num_violations": [0
],
"violation_rate": {"value": 0,
"data": [0
]
}
},
"property2": {"name": "string",
"num_violations": [0
],
"violation_rate": {"value": 0,
"data": [0
]
}
}
}
}
},
"hallucination": {"MODEL_RESPONSE": {"total": [0
],
"violation_rate": {"value": 0,
"data": [0
]
},
"average": {"value": 0,
"data": [0
]
}
}
},
"inputRelevance": {"MODEL_INPUT": {"entries": {"value": 0,
"data": [0
]
},
"violations": {"value": 0,
"data": [0
]
},
"violation_rate": {"value": 0,
"data": [0
]
}
}
},
"responseRelevance": {"MODEL_RESPONSE": {"entries": {"value": 0,
"data": [0
]
},
"violations": {"value": 0,
"data": [0
]
},
"contextRelevance": {"violations": {"value": 0,
"data": [0
]
},
"violation_rate": {"value": 0,
"data": [0
]
}
},
"responseRelevance": {"violations": {"value": 0,
"data": [0
]
},
"violation_rate": {"value": 0,
"data": [0
]
}
}
}
},
"latency": {"e2eLatency": {"percentile50": {"value": 0,
"data": [0
]
},
"percentile90": {"value": 0,
"data": [0
]
},
"percentile99": {"value": 0,
"data": [0
]
}
},
"inputGuardrailLatency": {"percentile50": {"value": 0,
"data": [0
]
},
"percentile90": {"value": 0,
"data": [0
]
},
"percentile99": {"value": 0,
"data": [0
]
}
},
"outputGuardrailLatency": {"percentile50": {"value": 0,
"data": [0
]
},
"percentile90": {"value": 0,
"data": [0
]
},
"percentile99": {"value": 0,
"data": [0
]
}
},
"llmCompletionLatency": {"percentile50": {"value": 0,
"data": [0
]
},
"percentile90": {"value": 0,
"data": [0
]
},
"percentile99": {"value": 0,
"data": [0
]
}
},
"requests": {"successful": {"value": 0,
"data": [0
]
},
"failed": {"value": 0,
"data": [0
]
}
},
"failureTypes": {"http4xx": {"value": 0,
"data": [0
]
},
"http5xx": {"value": 0,
"data": [0
]
}
}
},
"timestamps": [0
],
"rules": {"ruleViolations": 0,
"rulesActive": 0
}
}

DynamoGuard Chat

Guardrailed Chat

Apply policies to messages and send non-violating prompts to LLM. Routes messages to the LLM behind modelId. Blocks/modifies prompt/response as policies dictate

path Parameters

modelId required	string
session_id required	string Session id to link chat messages to

header Parameters

X-Num-Tokens	string Number of tokens analyzed by policies
Warning	string Warning displayed about tokens being not analyzed due to length

Request Body schema: application/json
required

clientId	string The client id of the sender
required	Array of objects (ChatMessage) OpenAI style messages to send to the LLM
metadata	object Metadata to be attached to the request
overridePolicyIds	Array of strings Explicit list of policies to apply instead of the ones attached to the model

Responses

Request samples

Payload

Content type

application/json

{"clientId": "string",
"messages": [{"role": "user",
"content": "string"
}
],
"metadata": { },
"overridePolicyIds": ["string"
]
}

Response samples

200

Content type

application/json

{"source": {"user": "string",
"session": "string",
"model": "string",
"client": "string"
},
"chat": {"queryId": "string",
"prompt": "string",
"preprocessedPrompt": "string",
"response": "string",
"desanitizedResponse": "string",
"postprocessedResponse": "string",
"ragContext": "string",
"error": "string"
},
"metadata": { },
"analyses": [{"text": "string",
"textType": "MODEL_INPUT",
"finalAction": "BLOCK",
"appliedPolicies": [{"policy": {"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
]
},
"outputs": { },
"action": "string"
}
],
"error": "string"
}
],
"usage": {"inputGuardrailTokens": 0,
"outputGuardrailTokens": 0,
"totalGuardrailTokens": 0
},
"warnings": [{"name": "Token Length Exceeded Warning",
"message": "string"
}
]
}

DynamoGuard Policy

Update policies applied to a model

path Parameters

id

required

string

Request Body schema: application/json
required

policyIds required	Array of strings List of policy ids to activate on the model
action required	string Enum: "ADD" "REMOVE" "REPLACE" Operation to perform on models

Responses

Request samples

Payload

Content type

application/json

{"policyIds": ["string"
],
"action": "ADD"
}

Create policy

Creates a policy on DynamoGuard. Each policy uses methodParams as arguments for the underlying policy and decisionParams to determine the action to take based on the policy result.

Request Body schema: application/json
required

name required	string Name of the policy
description	string Description of the policy
method required	string Enum: "PII" "TOXICITY" "ALIGNMENT" "HALLUCINATION" "RAG_HALLUCINATION" "INPUT_RELEVANCE" "RESPONSE_RELEVANCE" "OOB_HALLUCINATION" "KEYWORD" Type of policy Information about policy types
action required	string Default: "NONE" Enum: "BLOCK" "WARN" "REDACT" "SANITIZE" "NONE" The action to take if the policy is violated Information about policy actions
methodParams required	object The arguments for the policy. These differ depending on the policy type Information on policy types
decisionParams required	object The parameters for determining the policy action from outputs. These differ depending on the policy type Information on policy types
applicableTo required	string Default: "ALL" Enum: "INPUT" "OUTPUT" "ALL" Which type of message the policy should apply to
language	string Default: "en" Enum: "en" "ja" Language of the policy (ISO 639-1 code)

Responses

Request samples

Payload

Content type

application/json

{"name": "string",
"description": "string",
"method": "PII",
"action": "BLOCK",
"methodParams": { },
"decisionParams": { },
"applicableTo": "INPUT",
"language": "en"
}

Response samples

201

Content type

application/json

{"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"methodParams": { },
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
]
}

Get all policies

Retrieves all policies a user made. If modelId is specified, attaches policy statuses as well

query Parameters

policyType	string Enum: "PII" "TOXICITY" "ALIGNMENT" "HALLUCINATION" "RAG_HALLUCINATION" "INPUT_RELEVANCE" "RESPONSE_RELEVANCE" "OOB_HALLUCINATION" "KEYWORD" Policy type to filter on
modelId	string If specified, attaches policy status in relation to the modelId
expand	boolean If true, attaches dataset info. Only available for content policies
attachModels	boolean If specified, attaches models policy is applied to
inferred	boolean If true, returns inferred policy fields instead of the original policy
filterDeprecatedPolicies	boolean If true, filters out deprecated policies
page required	number
limit required	number
includeAllowedScopes required	boolean

Responses

Response samples

200

Content type

application/json

{"total": 0,
"policies": [{"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"methodParams": { },
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
],
"stage": "seed",
"feedback": {"promptId": "string",
"review": {"prompt": "string",
"isCompliant": true,
"feedback": true
},
"createdAt": "2019-08-24T14:15:22Z",
"creator": "string"
},
"trainingStatus": "CREATING",
"appliedStatus": "IN_DEVELOPMENT",
"resourceDeploymentStatus": "DEPLOYED",
"trainingStatusDetails": {"jobId": "string",
"progress": {"percentage": 0,
"totalNoOfItems": 0,
"noOfItemsProcessed": 0
},
"lastProgress": {"percentage": 0,
"totalNoOfItems": 0,
"noOfItemsProcessed": 0
},
"queuedForTooLong": true,
"wasResumed": true,
"failure": {"type": "INTERNAL_SERVER_ERROR",
"message": "string"
},
"totalElapsedTime": 0
},
"latestJobStat": { },
"models": ["string"
],
"creator": {"id": "string",
"email": "string"
}
}
]
}

Check if policy creation limit(s) reached

Responses

Response samples

200

Content type

application/json

{"alignment": {"hasReachedPolicyCreationLimit": true,
"currentNoOfPoliciesCreated": 0,
"totalNoOfPoliciesAllowed": 0
}
}

Update inferred policy

Updates only the inferred policy fields for ALIGNMENT type policies.

path Parameters

id

required

string

Policy id to update

Request Body schema: application/json
required

name required	string Name of the policy
description required	string Description of the policy
allowedBehaviors required	Array of strings List of allowed behaviors for the policy
disallowedBehaviors required	Array of strings List of disallowed behaviors for the policy

Responses

Request samples

Payload

Content type

application/json

{"name": "string",
"description": "string",
"allowedBehaviors": ["Provide factual information",
"Answer general knowledge questions"
],
"disallowedBehaviors": ["Share personal opinions",
"Provide medical advice"
]
}

Response samples

200

Content type

application/json

{"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"methodParams": { },
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
]
}

Update policy

Updates the base policy fields for all types of policies. Also updates the inferred policy fields for ALIGNMENT type policies.

path Parameters

id

required

string

Policy id to update

Request Body schema: application/json
required

name	string Name of the policy
description	string Description of the policy
action	string Enum: "BLOCK" "WARN" "REDACT" "SANITIZE" "NONE" The action to take if the policy is violated Information about policy actions
methodParams	object The arguments for the policy. These differ depending on the policy type Information on policy types
decisionParams	object The parameters for determining the policy action from outputs. These differ depending on the policy type Information on policy types
applicableTo	string Enum: "INPUT" "OUTPUT" "ALL" Which type of message the policy should apply to
language	string Enum: "en" "ja" Language of the policy (ISO 639-1 code)
isEnabled	boolean Default: true Whether the policy is enabled or disabled

Responses

Request samples

Payload

Content type

application/json

{"name": "string",
"description": "string",
"action": "BLOCK",
"methodParams": { },
"decisionParams": { },
"applicableTo": "INPUT",
"language": "en",
"isEnabled": true
}

Response samples

200

Content type

application/json

{"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"methodParams": { },
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
]
}

Get policy

Use query params to add additional information

path Parameters

id

required

string

Policy id to retrieve

query Parameters

expand	boolean If true, attaches dataset info. Only available for content policies
modelId	string If specified, attaches policy status in reference to the given model. Also filters policy stats.
promptId	string If specified, attaches feedback for the prompt
attachModels	boolean If specified, attaches models policy is applied to
inferred	boolean If true, returns inferred policy fields instead of the orignal policy

Responses

Response samples

200

Content type

application/json

{"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"methodParams": { },
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
],
"stage": "seed",
"feedback": {"promptId": "string",
"review": {"prompt": "string",
"isCompliant": true,
"feedback": true
},
"createdAt": "2019-08-24T14:15:22Z",
"creator": "string"
},
"trainingStatus": "CREATING",
"appliedStatus": "IN_DEVELOPMENT",
"resourceDeploymentStatus": "DEPLOYED",
"trainingStatusDetails": {"jobId": "string",
"progress": {"percentage": 0,
"totalNoOfItems": 0,
"noOfItemsProcessed": 0
},
"lastProgress": {"percentage": 0,
"totalNoOfItems": 0,
"noOfItemsProcessed": 0
},
"queuedForTooLong": true,
"wasResumed": true,
"failure": {"type": "INTERNAL_SERVER_ERROR",
"message": "string"
},
"totalElapsedTime": 0
},
"latestJobStat": { },
"models": ["string"
]
}

Delete policy

path Parameters

id

required

string

Policy id to delete

Responses

Response samples

200

Content type

application/json

{"failedToDeleteRelatedResources": ["guardrail",
"policy_access"
]
}

Add behaviors to policy

Appends allowed and disallowed behaviors to an existing alignment policy

path Parameters

id

required

string

Policy id to add behaviors to

Request Body schema: application/json
required

allowedBehaviors	Array of strings Additional allowed behaviors to add to the policy
disallowedBehaviors	Array of strings Additional disallowed behaviors to add to the policy

Responses

Request samples

Payload

Content type

application/json

{"allowedBehaviors": ["Provide factual information",
"Answer general knowledge questions"
],
"disallowedBehaviors": ["Share personal opinions",
"Provide medical advice"
]
}

Response samples

200

Content type

application/json

{"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"methodParams": { },
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
]
}

Generate behaviors for policy

Automatically generates allowed and disallowed behaviors for an alignment policy

path Parameters

id

required

string

Policy id to generate behaviors for

Responses

Add/remove policy to/from models

path Parameters

id

required

string

Request Body schema: application/json
required

modelIds required	Array of strings List of model ids to add/remove policy to
action required	string Enum: "ADD" "REMOVE" "REPLACE" Operation to perform on models

Responses

Request samples

Payload

Content type

application/json

{"modelIds": ["string"
],
"action": "ADD"
}

Add decision parameters to policy

Adds or updates decision parameters for a policy, starting with threshold configuration

path Parameters

id

required

string

Policy id to update decision parameters

Request Body schema: application/json
required

classification_threshold

required

number [ 0 .. 1 ]

Classification threshold value between 0 and 1

Responses

Request samples

Payload

Content type

application/json

{"classification_threshold": 1
}

Guardrail Dataset

Update guardrail dataset

Updates the configuration of a guardrail dataset

path Parameters

policyId

required

string

ID of the policy to update dataset for

Request Body schema: application/json
required

stage	string Stage of the guardrail dataset
	Array of objects (UpdateGuardrailDatapointDto) Array of datapoints for the guardrail dataset
bestTemperature	number Best temperature determined so far
bestInstruction	string Best instruction determined so far

Responses

Request samples

Payload

Content type

application/json

{"stage": "REVIEW",
"dataset": [{"id": "datapoint_123",
"round": 2,
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "generated",
"creatorId": "507f1f77bcf86cd799439011"
}
],
"bestTemperature": 0.7,
"bestInstruction": "You are an expert in providing helpful responses..."
}

Get guardrail dataset

Retrieves a guardrail dataset and all its metadata by policy ID

path Parameters

policyId

required

string

ID of the policy to get dataset for

query Parameters

page	number Default: 1 Page number for pagination (starting from 1)
limit	number Default: 300 Number of items per page. Takes default value of 300 if not specified. Maximum limit is 500
includeDataset	boolean Default: true Whether to include the dataset in the response
populateCreatorEmails	boolean Default: true Whether to populate creator emails in the response

Responses

Response samples

200

Content type

application/json

{"policyId": "policy_12345",
"creator": "admin",
"allowedBehaviors": ["Can talk about design principles",
"Can talk about generic programming questions"
],
"disallowedBehaviors": ["Cannot talk abouttax advise",
"Cannot talk about investment advise"
],
"stage": "seed",
"currentRound": 2,
"dataset": [{"id": "dp_123",
"round": 1,
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creatorId": "user_abc",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"datasetPageDetails": {"page": 1,
"limit": 100,
"totalPages": 28,
"totalItems": 2750
},
"feedback": [{"_id": "507f1f77bcf86cd799439011",
"promptId": "prompt_123",
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creator": "507f1f77bcf86cd799439012",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"version": 3,
"bestTemperature": 0.8,
"bestInstruction": "You are an expert in providing helpful responses...",
"created_at": "2024-03-14T12:00:00Z",
"updated_at": "2024-03-14T12:00:00Z"
}

Create input guardrail dataset

Creates a new input guardrail dataset for a policy

path Parameters

policyId

required

string

ID of the policy to create dataset for

Request Body schema: application/json
required

allowedBehaviors	Array of strings An array of allowed behaviors
disallowedBehaviors	Array of strings An array of disallowed behaviors
manualCreation required	boolean Indicates whether this dataset was manually created

Responses

Request samples

Payload

Content type

application/json

{"allowedBehaviors": ["Can talk about generic programming questions.",
"Can talk about design principles."
],
"disallowedBehaviors": ["Cannot talk about tax advise.",
"Cannot talk about investment advise."
],
"manualCreation": false
}

Response samples

201

Content type

application/json

{"policyId": "policy_12345",
"creator": "admin",
"allowedBehaviors": ["Can talk about design principles",
"Can talk about generic programming questions"
],
"disallowedBehaviors": ["Cannot talk abouttax advise",
"Cannot talk about investment advise"
],
"stage": "seed",
"currentRound": 2,
"dataset": [{"id": "dp_123",
"round": 1,
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creatorId": "user_abc",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"datasetPageDetails": {"page": 1,
"limit": 100,
"totalPages": 28,
"totalItems": 2750
},
"feedback": [{"_id": "507f1f77bcf86cd799439011",
"promptId": "prompt_123",
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creator": "507f1f77bcf86cd799439012",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"version": 3,
"bestTemperature": 0.8,
"bestInstruction": "You are an expert in providing helpful responses...",
"created_at": "2024-03-14T12:00:00Z",
"updated_at": "2024-03-14T12:00:00Z"
}

Create output guardrail dataset

Creates a new output guardrail dataset for a policy

path Parameters

policyId

required

string

ID of the policy to create dataset for

Request Body schema: application/json
required

allowedBehaviors	Array of strings An array of allowed behaviors
disallowedBehaviors	Array of strings An array of disallowed behaviors
manualCreation required	boolean Indicates whether this dataset was manually created

Responses

Request samples

Payload

Content type

application/json

{"allowedBehaviors": ["Can talk about generic programming questions.",
"Can talk about design principles."
],
"disallowedBehaviors": ["Cannot talk about tax advise.",
"Cannot talk about investment advise."
],
"manualCreation": false
}

Response samples

201

Content type

application/json

{"policyId": "policy_12345",
"creator": "admin",
"allowedBehaviors": ["Can talk about design principles",
"Can talk about generic programming questions"
],
"disallowedBehaviors": ["Cannot talk abouttax advise",
"Cannot talk about investment advise"
],
"stage": "seed",
"currentRound": 2,
"dataset": [{"id": "dp_123",
"round": 1,
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creatorId": "user_abc",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"datasetPageDetails": {"page": 1,
"limit": 100,
"totalPages": 28,
"totalItems": 2750
},
"feedback": [{"_id": "507f1f77bcf86cd799439011",
"promptId": "prompt_123",
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creator": "507f1f77bcf86cd799439012",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"version": 3,
"bestTemperature": 0.8,
"bestInstruction": "You are an expert in providing helpful responses...",
"created_at": "2024-03-14T12:00:00Z",
"updated_at": "2024-03-14T12:00:00Z"
}

Review input guardrail dataset

Reviews and updates an input guardrail dataset

path Parameters

policyId

required

string

ID of the policy to review dataset for

Request Body schema: application/json
required

required

Array of objects (ReviewedInputGuardrailDataset)

Array of reviewed guardrail datapoints for input

Array

isAccepted required	boolean Whether the datapoint is accepted by the user. False means it was discarded.
prompt	string The prompt string of the datapoint
isCompliant	boolean Indicates if the prompt is compliant with the policy
feedback	string Any user feedback on the prompt
id required	string Unique identifier for the datapoint

Responses

Request samples

Payload

Content type

application/json

{"reviewedDataset": [{"isAccepted": true,
"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"feedback": "The prompt seems compliant and safe for our guardrail.",
"id": "abc123"
}
]
}

Response samples

200

Content type

application/json

{"policyId": "policy_12345",
"creator": "admin",
"allowedBehaviors": ["Can talk about design principles",
"Can talk about generic programming questions"
],
"disallowedBehaviors": ["Cannot talk abouttax advise",
"Cannot talk about investment advise"
],
"stage": "seed",
"currentRound": 2,
"dataset": [{"id": "dp_123",
"round": 1,
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creatorId": "user_abc",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"datasetPageDetails": {"page": 1,
"limit": 100,
"totalPages": 28,
"totalItems": 2750
},
"feedback": [{"_id": "507f1f77bcf86cd799439011",
"promptId": "prompt_123",
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creator": "507f1f77bcf86cd799439012",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"version": 3,
"bestTemperature": 0.8,
"bestInstruction": "You are an expert in providing helpful responses...",
"created_at": "2024-03-14T12:00:00Z",
"updated_at": "2024-03-14T12:00:00Z"
}

Review output guardrail dataset

Reviews and updates an output guardrail dataset

path Parameters

policyId

required

string

ID of the policy to review dataset for

Request Body schema: application/json
required

required

Array of objects (ReviewedOutputGuardrailDataset)

Array of reviewed guardrail datapoints for output

Array

isAccepted required	boolean Whether the datapoint is accepted by the user. False means it was discarded.
prompt	string The prompt string of the datapoint
isCompliant	boolean Indicates if the prompt is compliant with the policy
feedback	string Any user feedback on the prompt
id required	string Unique identifier for the datapoint
response	string The response string of the datapoint
userName	string Name of the user who reviewed the datapoint
userId	string User ID of the reviewer
updateTime	string Timestamp of the review update (string format)

Responses

Request samples

Payload

Content type

application/json

{"reviewedDataset": [{"isAccepted": true,
"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"feedback": "The prompt seems compliant and safe for our guardrail.",
"id": "def456",
"response": "42 is the meaning of life.",
"userName": "Alice",
"userId": "user_12345",
"updateTime": "2024-03-14T12:00:00Z"
}
]
}

Response samples

200

Content type

application/json

{"policyId": "policy_12345",
"creator": "admin",
"allowedBehaviors": ["Can talk about design principles",
"Can talk about generic programming questions"
],
"disallowedBehaviors": ["Cannot talk abouttax advise",
"Cannot talk about investment advise"
],
"stage": "seed",
"currentRound": 2,
"dataset": [{"id": "dp_123",
"round": 1,
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creatorId": "user_abc",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"datasetPageDetails": {"page": 1,
"limit": 100,
"totalPages": 28,
"totalItems": 2750
},
"feedback": [{"_id": "507f1f77bcf86cd799439011",
"promptId": "prompt_123",
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creator": "507f1f77bcf86cd799439012",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"version": 3,
"bestTemperature": 0.8,
"bestInstruction": "You are an expert in providing helpful responses...",
"created_at": "2024-03-14T12:00:00Z",
"updated_at": "2024-03-14T12:00:00Z"
}

Feedback guardrail dataset

Provides feedback on a guardrail dataset

path Parameters

policyId

required

string

ID of the policy to provide feedback for

Responses

Response samples

200

Content type

application/json

{"policyId": "policy_12345",
"creator": "admin",
"allowedBehaviors": ["Can talk about design principles",
"Can talk about generic programming questions"
],
"disallowedBehaviors": ["Cannot talk abouttax advise",
"Cannot talk about investment advise"
],
"stage": "seed",
"currentRound": 2,
"dataset": [{"id": "dp_123",
"round": 1,
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creatorId": "user_abc",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"datasetPageDetails": {"page": 1,
"limit": 100,
"totalPages": 28,
"totalItems": 2750
},
"feedback": [{"_id": "507f1f77bcf86cd799439011",
"promptId": "prompt_123",
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creator": "507f1f77bcf86cd799439012",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}
],
"version": 3,
"bestTemperature": 0.8,
"bestInstruction": "You are an expert in providing helpful responses...",
"created_at": "2024-03-14T12:00:00Z",
"updated_at": "2024-03-14T12:00:00Z"
}

Get feedback guardrail dataset

Retrieves feedback on a guardrail dataset

path Parameters

policyId

required

string

ID of the policy to get feedback for

Responses

Response samples

200

Content type

application/json

{"_id": "507f1f77bcf86cd799439011",
"promptId": "prompt_123",
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "user",
"creator": "507f1f77bcf86cd799439012",
"createdAt": "2024-03-14T12:00:00Z",
"updatedAt": "2024-03-14T12:00:00Z"
}

Apply guardrail feedback

Applies feedback on a guardrail dataset

path Parameters

policyId

required

string

ID of the policy to apply feedback for

Request Body schema: application/json
required

ids

required

Array of strings

Array of feedback item IDs to apply

Responses

Request samples

Payload

Content type

application/json

{"ids": ["feedback_111",
"feedback_222"
]
}

Get guardrail dataset

Retrieves a guardrail dataset by policy ID

path Parameters

policyId

required

string

ID of the policy to get dataset for

Request Body schema: application/json
required

ids

required

Array of strings

Array of feedback item IDs to reject

Responses

Request samples

Payload

Content type

application/json

{"ids": ["feedback_333",
"feedback_444"
]
}

Get latest finetuned model info

Retrieves the latest finetuned model info for a policy

path Parameters

policyId

required

string

ID of the policy to get latest finetuned model info for

Responses

Response samples

200

Content type

application/json

{"status": "COMPLETED",
"deploymentId": "no-political-advice-09809a0c098efb0",
"creator": "user_123",
"policyIds": ["68ac89f890a89f890a89f890"
],
"modelType": "input_mini",
"fineTunedModelZipS3Path": "s3://my-bucket/models/model.zip",
"isModelEncrypted": true,
"trainingJobId": "job_123",
"errorMessage": "string"
}

Get fine-tuned model info

Retrieves fine-tuned model info by model ID

path Parameters

modelId

required

string

ID of the fine-tuned model to get info for

Responses

Response samples

200

Content type

application/json

{"status": "COMPLETED",
"modelId": "finetuned_model_abc",
"policyIds": ["policy_123",
"policy_456"
],
"deploymentId": "no-political-advice-09809a0c098efb0"
}

Update fine-tuned model info

Updates the fine-tuned model info by model ID (also deploys the model if needed)

path Parameters

modelId

required

string

ID of the fine-tuned model to update info for

Request Body schema: application/json
required

status required	string Enum: "QUEUED" "TRAINING" "TRAINED" "DEPLOYING" "COMPLETED" "DEPLOYMENT_FAILED" "FAILED" "CANCELLED" Status of the fine-tuning process
datasetS3Path	string Path to the dataset in S3
fineTunedModelZipS3Path	string Path to the fine-tuned model zip in S3
isModelEncrypted	boolean Is the model encrypted
error	string Error message if training fails

Responses

Request samples

Payload

Content type

application/json

{"status": "QUEUED",
"datasetS3Path": "s3://my-bucket/datasets/dataset.csv",
"fineTunedModelZipS3Path": "s3://my-bucket/models/model.zip",
"isModelEncrypted": true,
"error": "Out of memory error"
}

Response samples

200

Content type

application/json

{"status": "COMPLETED",
"modelId": "finetuned_model_abc",
"policyIds": ["policy_123",
"policy_456"
],
"deploymentId": "no-political-advice-09809a0c098efb0"
}

Get deployment status of a fine-tuned model

Retrieves the deployment status of a fine-tuned model

path Parameters

modelId

required

string

ID of the fine-tuned model to get deployment status for

Responses

Response samples

200

Content type

application/json

{"status": "COMPLETED",
"reason": "DeploymentReady",
"message": "string",
"readyReplicas": 0,
"observedGeneration": 0,
"modelPhase": "string"
}

Redeploy a policy

Attempts to redeploy a policy

path Parameters

policyId

required

string

ID of the policy to be redeployed

Responses

Response samples

200

Content type

application/json

{"status": "COMPLETED",
"modelId": "finetuned_model_abc",
"policyIds": ["policy_123",
"policy_456"
],
"deploymentId": "no-political-advice-09809a0c098efb0"
}

Update guardrail config

Updates the configuration of a guardrail policy

path Parameters

policyId

required

string

ID of the policy to update config for

Request Body schema: application/json
required

name	string Name of the guardrail config
description	string Description of the guardrail config
allowedBehaviors	Array of strings List of allowed behaviors
disallowedBehaviors	Array of strings List of disallowed behaviors

Responses

Request samples

Payload

Content type

application/json

{"name": "No tax/investment advise",
"description": "This guardrail prevents tax and investment advise.",
"allowedBehaviors": ["Can talk about generic programming questions",
"Can talk about design principles"
],
"disallowedBehaviors": ["Cannot talk about tax advise",
"Cannot talk about investment advise"
]
}

Response samples

200

Content type

application/json

{"name": "No tax/investment advise",
"description": "This guardrail prevents tax and investment advise.",
"allowedBehaviors": ["Can talk about generic programming questions",
"Can talk about design principles"
],
"disallowedBehaviors": ["Cannot talk about tax advise",
"Cannot talk about investment advise"
],
"_id": "507f1f77bcf86cd799439011",
"policyId": "507f1f77bcf86cd799439012",
"version": 1,
"created_at": "2024-03-14T12:00:00Z",
"updated_at": "2024-03-14T12:00:00Z"
}

Add generated guardrail dataset points

Adds generated guardrail dataset points for a given policy

path Parameters

policyId

required

string

ID of the policy to add dataset points for

Request Body schema: application/json
required

required

Array of objects (UpdateGuardrailDatapointDto)

An array of datapoints to add to the guardrail dataset

Array

id required	string Unique identifier for the datapoint
round required	number Round number for the datapoint
required	object Generated data for this datapoint
	object Optional review data for this datapoint
creatorType	string Enum: "user" "model" "file-upload" "generated" "feedback" Type of creator for this datapoint
creatorId	string MongoDB ObjectId of the creator

Responses

Request samples

Payload

Content type

application/json

{"dataset": [{"id": "datapoint_123",
"round": 2,
"generated": {"response": "Factory design pattern is a creational pattern...",
"isCompliant": true,
"prompt": "Can you please tell me what factory design pattern is?",
"rationale": "Prompt is asking for a design pattern explanation, which is safe.",
"trickiness": "in_domain",
"relevantBehaviors": ["Can talk about design principles"
],
"violatedCategoriesResolutionCodes": ["category_123"
],
"policyAnalysisRationale": "This response is safe"
},
"review": {"prompt": "Can you please tell me what factory design pattern is?",
"isCompliant": true,
"isAccepted": true,
"feedback": "Prompt is asking for a design pattern explanation and the response answers it, which is safe.",
"response": "Factory design pattern is a creational pattern..."
},
"creatorType": "generated",
"creatorId": "507f1f77bcf86cd799439011"
}
]
}

Update guardrail dataset point

Updates a single guardrail dataset point for a given policy

path Parameters

policyId

required

string

ID of the policy to update dataset point for

Request Body schema: application/json
required

id required	string Unique identifier for the dataset point
prompt required	string The prompt text for the dataset point
response	string In case of output guard-rails, response for the dataset point
isAccepted	boolean Indicates if the datapoint is accepted
isCompliant	boolean Indicates if the datapoint is compliant

Responses

Request samples

Payload

Content type

application/json

{"id": "xyz789",
"prompt": "Can you please tell me what factory design pattern is?",
"response": "Factory design pattern is a creational pattern... ",
"isAccepted": true,
"isCompliant": true
}

Delete guardrail dataset points

Deletes guardrail dataset points for a given policy

path Parameters

policyId

required

string

ID of the policy to delete dataset points for

Request Body schema: application/json
required

ids

required

Array of strings

Array of IDs to delete

Responses

Request samples

Payload

Content type

application/json

{"ids": ["xyz789",
"abc456"
]
}

Add user-defined guardrail dataset point

Adds a single user-defined guardrail dataset point for a given policy

path Parameters

policyId

required

string

ID of the policy to add dataset point for

Request Body schema: application/json
required

prompt required	string Prompt text for the user-defined dataset point
response	string In case of output guard-rails, the response for the prompt
isCompliant required	boolean Indicates if the datapoint is compliant

Responses

Request samples

Payload

Content type

application/json

{"prompt": "Can you please tell me what factory design pattern is?",
"response": "Factory design pattern is a creational pattern... ",
"isCompliant": true
}

Get pending changes

Retrieves pending changes for a guardrail dataset, which are the diffs that aren't yet trained.

path Parameters

policyId

required

string

Responses

Response samples

200

Content type

application/json

{"diffs": [{"_id": "507f1f77bcf86cd799439011",
"currentVersion": {"grdVersion": "v2",
"grdConfigVersion": "v1.1"
},
"previousVersion": {"grdVersion": "v2",
"grdConfigVersion": "v1.1"
},
"policyId": "policy_abc",
"diff": [{"type": "behaviors",
"additions": 5,
"deletions": 2,
"updations": 1,
"restoredFrom": {"id": "507f1f77bcf86cd799439011",
"title": "Version 1.0"
}
}
],
"status": "in_training",
"creator": {"email": "user@example.com"
},
"createdAt": "2025-01-01T12:00:00Z",
"updatedAt": "2025-01-02T15:00:00Z",
"isInitalChange": false,
"isPreApproveReadyChange": false
}
],
"shouldRegenerate": true
}

Start training

Starts training for a guardrail dataset

path Parameters

policyId

required

string

ID of the policy to train

Request Body schema: application/json
required

title	string Optional title for the current model training session
description required	string Description of the training session

Responses

Request samples

Payload

Content type

application/json

{"title": "Training session 1",
"description": "This training focuses on moderate content response behaviors."
}

Response samples

200

Content type

application/json

{"diffs": [{"_id": "507f1f77bcf86cd799439011",
"currentVersion": {"grdVersion": "v2",
"grdConfigVersion": "v1.1"
},
"previousVersion": {"grdVersion": "v2",
"grdConfigVersion": "v1.1"
},
"policyId": "policy_abc",
"diff": [{"type": "behaviors",
"additions": 5,
"deletions": 2,
"updations": 1,
"restoredFrom": {"id": "507f1f77bcf86cd799439011",
"title": "Version 1.0"
}
}
],
"status": "in_training",
"creator": {"email": "user@example.com"
},
"createdAt": "2025-01-01T12:00:00Z",
"updatedAt": "2025-01-02T15:00:00Z",
"isInitalChange": false,
"isPreApproveReadyChange": false
}
],
"shouldRegenerate": true
}

Restart training

Restarts training for a guardrail dataset, marking all AWAITING_TRAINING diffs as IN_TRAINING and restarting the training job

path Parameters

policyId

required

string

ID of the policy to restart training

Responses

Response samples

200

Content type

application/json

{"diffs": [{"_id": "507f1f77bcf86cd799439011",
"currentVersion": {"grdVersion": "v2",
"grdConfigVersion": "v1.1"
},
"previousVersion": {"grdVersion": "v2",
"grdConfigVersion": "v1.1"
},
"policyId": "policy_abc",
"diff": [{"type": "behaviors",
"additions": 5,
"deletions": 2,
"updations": 1,
"restoredFrom": {"id": "507f1f77bcf86cd799439011",
"title": "Version 1.0"
}
}
],
"status": "in_training",
"creator": {"email": "user@example.com"
},
"createdAt": "2025-01-01T12:00:00Z",
"updatedAt": "2025-01-02T15:00:00Z",
"isInitalChange": false,
"isPreApproveReadyChange": false
}
],
"shouldRegenerate": true
}

Cancel training

Cancels training for a guardrail dataset

path Parameters

policyId

required

string

ID of the policy to cancel training

Responses

Response samples

200

Content type

application/json

{"diffs": [{"_id": "507f1f77bcf86cd799439011",
"currentVersion": {"grdVersion": "v2",
"grdConfigVersion": "v1.1"
},
"previousVersion": {"grdVersion": "v2",
"grdConfigVersion": "v1.1"
},
"policyId": "policy_abc",
"diff": [{"type": "behaviors",
"additions": 5,
"deletions": 2,
"updations": 1,
"restoredFrom": {"id": "507f1f77bcf86cd799439011",
"title": "Version 1.0"
}
}
],
"status": "in_training",
"creator": {"email": "user@example.com"
},
"createdAt": "2025-01-01T12:00:00Z",
"updatedAt": "2025-01-02T15:00:00Z",
"isInitalChange": false,
"isPreApproveReadyChange": false
}
],
"shouldRegenerate": true
}

Cancel data generation

Cancels data generation for a guardrail dataset

path Parameters

policyId

required

string

ID of the policy to cancel data generation for

Responses

Restart data generation

Restarts data generation for a guardrail dataset from scratch, discarding all non-user reviewed datapoints

path Parameters

policyId

required

string

ID of the policy to restart data generation for

Responses

Resume data generation

Resumes data generation for a guardrail dataset from where it was previously cancelled, failed, or timed out

path Parameters

policyId

required

string

ID of the policy to resume data generation for

Responses

Update training status

Updates the training status for a guardrail dataset

path Parameters

policyId

required

string

ID of the policy to update training status

Request Body schema: application/json
required

fineTunedModeId required	string ID of the fine-tuned model
isSuccess required	boolean Indicates if the training was successful
bestInstruction	string Best instruction found during training
bestTemperature	number Best temperature found during training

Responses

Request samples

Payload

Content type

application/json

{"fineTunedModeId": "finetuned_007",
"isSuccess": true,
"bestInstruction": "You are an expert in providing helpful responses...",
"bestTemperature": 0.7
}

Get versioning info

Retrieves versioning info for a guardrail dataset

path Parameters

policyId

required

string

ID of the policy to get versioning info

Responses

Response samples

200

Content type

application/json

[{"policyId": "507f1f77bcf86cd799439011",
"diffs": [{"_id": "507f1f77bcf86cd799439011",
"currentVersion": {"grdVersion": "v2",
"grdConfigVersion": "v1.1"
},
"previousVersion": {"grdVersion": "v2",
"grdConfigVersion": "v1.1"
},
"policyId": "policy_abc",
"diff": [{"type": "behaviors",
"additions": 5,
"deletions": 2,
"updations": 1,
"restoredFrom": {"id": "507f1f77bcf86cd799439011",
"title": "Version 1.0"
}
}
],
"status": "in_training",
"creator": {"email": "user@example.com"
},
"createdAt": "2025-01-01T12:00:00Z",
"updatedAt": "2025-01-02T15:00:00Z",
"isInitalChange": false,
"isPreApproveReadyChange": false
}
],
"versionInfo": {"id": "507f1f77bcf86cd799439011",
"title": "Minor changes on policy_xxx",
"description": "Fixed mistakes and updated instructions.",
"creator": {"email": "user@example.com"
},
"createdAt": "2025-01-01T10:00:00Z"
}
}
]

Restore version

Restores a guardrail dataset to a specific version

path Parameters

policyId

required

string

ID of the policy to revert

Request Body schema: application/json
required

versionId

required

string

ID of the version to revert to

Responses

Request samples

Payload

Content type

application/json

{"versionId": "version_123"
}

Regenerate guardrail dataset

Regenerates the guardrail dataset for a policy

path Parameters

policyId

required

string

ID of the policy to regenerate dataset

Responses

Upload CSV dataset

Upload a CSV file to populate the guardrail dataset

path Parameters

policyId

required

string

ID of the policy to upload dataset

Request Body schema: multipart/form-data
required

file	string <binary>
replaceDuplicates	boolean

Responses

Health

API Health Endpoint

To check if the api is healthy

Responses

Version

API Version Endpoint

To return the api version

Responses

DynamoGuard Extension Get Domains

ExtensionController_getDomains

Responses

Response samples

200

Content type

application/json

["string"
]

Application

Create a new application

Create a new application with optional AI systems

Request Body schema: application/json
required

name required	string Name of the application
tag	string Tag/description of the application
aiSystems	Array of strings IDs of AI systems to associate with the application

Responses

Request samples

Payload

Content type

application/json

{"name": "Chat Application",
"tag": "AI-powered chat application for customer support",
"aiSystems": ["6078f9b8b328f40068b9fd1b",
"6078f9b8b328f40068b9fd1c"
]
}

Response samples

201

Content type

application/json

{"id": "123",
"name": "My Awesome App",
"tag": "v1.0",
"createdAt": "2019-08-24T14:15:22Z",
"createdBy": { },
"updatedAt": "2019-08-24T14:15:22Z",
"metadata": { }
}

List applications with pagination

query Parameters

page required	number
limit required	number
includeAllowedScopes required	boolean
includeModerationPolicies required	boolean
includePolicyCreator required	boolean
includeApplicationCreator required	boolean
includePolicyStatus required	boolean

Responses

Response samples

200

Content type

application/json

{ }

Share application with collaborator

Share a application with a collaborator and assign specific roles

path Parameters

_id

required

string

Request Body schema: application/json
required

userEmail required	string Email of the collaborator
resourceLevelRoles required	Array of strings Role of the collaborator at which this resource is being shared

Responses

Request samples

Payload

Content type

application/json

{"userEmail": "string",
"resourceLevelRoles": ["string"
]
}

Remove collaborator from application

Remove a collaborator and their associated roles from the application

path Parameters

_id

required

string

Request Body schema: application/json
required

userEmail

required

string

Email of the collaborator

Responses

Request samples

Payload

Content type

application/json

{"userEmail": "string"
}

Update application public access

Update the public access settings and roles for the application

path Parameters

_id

required

string

Request Body schema: application/json
required

isPublic required	boolean Whether the resource is shared with public
publicResourceLevelRoles	Array of strings The role at which the public is allowed to access this resource

Responses

Request samples

Payload

Content type

application/json

{"isPublic": true,
"publicResourceLevelRoles": ["string"
]
}

Get application access information

Get the resource level access information for the application

path Parameters

_id

required

string

Responses

Get application resource level roles

Get the resource level roles associated with the application

Responses

Get application resource level scopes

Get the resource level scopes associated with the application for the current user

path Parameters

_id

required

string

Responses

Get a specific application by its ID

path Parameters

id

required

number

The ID of the application to retrieve

query Parameters

includeAllowedScopes required	boolean
includeModerationPolicies required	boolean
includePolicyCreator required	boolean
includePolicyStatus required	boolean

Responses

Response samples

200

Content type

application/json

{ }

Update an existing application

Updates specified fields of an existing application.

path Parameters

id

required

integer

Application ID

Request Body schema: application/json
required

name	string [ 1 .. 255 ] characters The updated name of the application.
tag	string The updated tag or short description for the application.
metadata	object Optional metadata for the application, as a JSON object.

Responses

Request samples

Payload

Content type

application/json

{"name": "Enhanced Chat Application",
"tag": "Updated customer support chat interface",
"metadata": {"status": "draft",
"version": "2.0"
}
}

Response samples

200

Content type

application/json

{"id": "123",
"name": "My Awesome App",
"tag": "v1.0",
"createdAt": "2019-08-24T14:15:22Z",
"createdBy": { },
"updatedAt": "2019-08-24T14:15:22Z",
"metadata": { }
}

Delete an existing application

Deletes an existing application by its ID.

path Parameters

id

required

integer

Application ID

Responses

Response samples

200

Content type

application/json

{"deletedApplicationId": "6078f9b8b328f40068b9fd2a"
}

Get all AI systems for application (active + removed)

Retrieves all AI systems that were ever linked to the application for logs filtering UI

path Parameters

id

required

number

The ID of the application to retrieve AI systems for

Responses

Response samples

200

Content type

application/json

{ }

Get application logs

Retrieve logs for a specific application with enhanced AI system information and filtering capabilities.

path Parameters

id

required

number

The ID of the application

query Parameters

startTime	number The start time (in UTC seconds) to filter by
endTime	number The end time (in UTC seconds) to filter by
pageNum	number Default: 1 The page number
perPage	number Default: 50 The number of results per page
sortDir	string Default: "desc" Enum: "asc" "desc" The direction to sort results by.
policyId	string The policy id to filter by
policyType	string Enum: "PII" "TOXICITY" "ALIGNMENT" "HALLUCINATION" "RAG_HALLUCINATION" "INPUT_RELEVANCE" "RESPONSE_RELEVANCE" "OOB_HALLUCINATION" "KEYWORD" The policy type to filter by
actions	Array of strings List of actions to filter by
textType	string Enum: "MODEL_INPUT" "MODEL_RESPONSE" The text type to filter by
action	string Enum: "BLOCK" "WARN" "REDACT" "SANITIZE" "NONE" The policy action to filter by
clientId	string The client id to filter by
userId	string The user id to filter by
policyTypes	Array of strings List of policy types to filter by
metadata	string Example: metadata={origin: {platform: "chatgpt.com", browser: "Chrome"}, version: "3.16"} User-defined log metadata to filter by. Should be a json string
aiSystemIds	Array of strings Example: aiSystemIds=507f1f77bcf86cd799439011&aiSystemIds=507f1f77bcf86cd799439012 Array of specific AI System IDs to filter logs
includeRemovedAiSystems	boolean Default: true Example: includeRemovedAiSystems=true Whether to include logs from AI systems that have been removed from the application

Responses

Response samples

200

Content type

application/json

{"logs": [{"timestamp": "string",
"analyses": [{"text": "string",
"textType": "MODEL_INPUT",
"appliedPolicies": [{"outputs": { },
"violated": true,
"feedback": { },
"promptId": "string",
"policy": {"id": "string",
"name": "string",
"description": "string",
"isEnabled": true,
"method": "PII",
"action": "BLOCK",
"methodParams": { },
"decisionParams": { },
"language": "en",
"applicableTo": "INPUT",
"created_at": "2019-08-24T14:15:22Z",
"creatorId": "string",
"updated_at": "2019-08-24T14:15:22Z",
"usecaseId": "string",
"allowedBehaviors": ["string"
],
"disallowedBehaviors": ["string"
]
},
"action": "BLOCK"
}
]
}
],
"action": "BLOCK",
"metadata": { },
"applicationAiSystemId": "507f1f77bcf86cd799439011",
"applicationAiSystemName": "GPT-4 Content Moderator",
"applicationAiSystemRemovedAt": "2024-01-15T10:30:00.000Z"
}
],
"totalCount": 150,
"warnings": [{"name": "Token Length Exceeded Warning",
"message": "string"
}
]
}

Get application logs count

Retrieve the count of logs for a specific application with the same filtering capabilities as the main logs endpoint.

path Parameters

id

required

number

The ID of the application

query Parameters

startTime	number The start time (in UTC seconds) to filter by
endTime	number The end time (in UTC seconds) to filter by
pageNum	number Default: 1 The page number
perPage	number Default: 50 The number of results per page
sortDir	string Default: "desc" Enum: "asc" "desc" The direction to sort results by.
policyId	string The policy id to filter by
policyType	string Enum: "PII" "TOXICITY" "ALIGNMENT" "HALLUCINATION" "RAG_HALLUCINATION" "INPUT_RELEVANCE" "RESPONSE_RELEVANCE" "OOB_HALLUCINATION" "KEYWORD" The policy type to filter by
actions	Array of strings List of actions to filter by
textType	string Enum: "MODEL_INPUT" "MODEL_RESPONSE" The text type to filter by
action	string Enum: "BLOCK" "WARN" "REDACT" "SANITIZE" "NONE" The policy action to filter by
clientId	string The client id to filter by
userId	string The user id to filter by
policyTypes	Array of strings List of policy types to filter by
metadata	string Example: metadata={origin: {platform: "chatgpt.com", browser: "Chrome"}, version: "3.16"} User-defined log metadata to filter by. Should be a json string
aiSystemIds	Array of strings Example: aiSystemIds=507f1f77bcf86cd799439011&aiSystemIds=507f1f77bcf86cd799439012 Array of specific AI System IDs to filter logs
includeRemovedAiSystems	boolean Default: true Example: includeRemovedAiSystems=true Whether to include logs from AI systems that have been removed from the application

Responses

Response samples

200

Content type

application/json

0

0

Assign (add/remove) AI Systems to/from an Application

Adds and/or removes AI Systems from an application and synchronizes moderation policies.

path Parameters

id

required

number

The ID of the application

Request Body schema: application/json
required

addAiSystems	Array of strings Array of MongoDB MLModel IDs to add to the application.
removeAiSystems	Array of strings Array of MongoDB MLModel IDs to remove from the application.

Responses

Request samples

Payload

Content type

application/json

{"addAiSystems": ["6078f9b8b328f40068b9fd1d",
"6078f9b8b328f40068b9fd2e"
],
"removeAiSystems": ["6078f9b8b328f40068b9fd1d",
"6078f9b8b328f40068b9fd2e"
]
}

Response samples

200

Content type

application/json

{"applicationId": "123",
"processedResults": [{"aiSystemId": "6078f9b8b328f40068b9fd1d",
"operationType": "ADD",
"operationStatus": "SUCCESS_ADDED_POLICIES_SYNCED",
"status": true,
"reason": "string",
"linkedAt": "2025-05-19T08:28:17.406Z",
"removedAt": "2025-05-19T08:28:17.406Z"
}
]
}

Assign (add/remove) Policies to/from an Application

Adds and/or removes Policies from an application and synchronizes them with all existing AI systems.

path Parameters

id

required

number

The ID of the application

Request Body schema: application/json
required

addPolicies	Array of strings Array of MongoDB Policy IDs to add to the application.
removePolicies	Array of strings Array of MongoDB Policy IDs to remove from the application.

Responses

Request samples

Payload

Content type

application/json

{"addPolicies": ["6078f9b8b328f40068b9fd1d",
"6078f9b8b328f40068b9fd2e"
],
"removePolicies": ["6078f9b8b328f40068b9fd1d",
"6078f9b8b328f40068b9fd2e"
]
}

Response samples

200

Content type

application/json

{"applicationId": "123",
"processedResults": [{"policyId": "6078f9b8b328f40068b9fd1d",
"operationType": "ADD",
"operationStatus": "SUCCESS_ADDED_POLICIES_SYNCED",
"status": true,
"reason": "Successfully linked and policies synchronized.",
"linkedAt": "2025-05-19T08:28:17.406Z",
"removedAt": "2025-05-19T08:28:17.406Z"
}
]
}

Manage Policies for an Application

Replace existing policies with specific sets of policies for individual AI systems within the application.

path Parameters

id

required

number

The ID of the application

Request Body schema: application/json
required

object (ManagePoliciesDto)

Responses

Request samples

Payload

Content type

application/json

{ }

Response samples

207

Content type

application/json

{ }

DynamoFL API (3.24.3)

Users

Get current user

Responses

Create user

Request Body schema: application/jsonrequired

Responses

Request samples

Response samples

Get current user with Keycloak info

Responses

Reset password for non-onboarded user

Request Body schema: application/jsonrequired

Responses

Request samples

Response samples

Reset password for onboarded user

Request Body schema: application/jsonrequired

Responses

Request samples

Get all users

query Parameters

Responses

Response samples

Update organization level role

path Parameters

Request Body schema: application/jsonrequired

Responses

Request samples

Delete user

path Parameters

Responses

Get current user organization level scopes

Responses

Get current user terms consent status

Responses

Response samples

Update current user terms consent

Request Body schema: application/jsonrequired

Responses

Request samples

Create API token

Responses

Response samples

TermsConsentController_getAllTermsConsents

Responses

ModerationController_generatePolicyBehaviors

Request Body schema: application/jsonrequired

Responses

Request samples

PolicyController_getPolicyResourceLevelRoles

Responses

PolicyController_addUseCase

Request Body schema: application/jsonrequired

Responses

Request samples

PolicyController_getUseCases

Responses

PolicyController_updateUsecase

path Parameters

Request Body schema: application/jsonrequired

Responses

Request samples

PolicyController_getUsecase

path Parameters

Responses

PolicyController_deleteUsecase

path Parameters

Responses

PolicyController_getAnyUsecase

path Parameters

Responses

InClusterModelController_upsertModel

path Parameters

Request Body schema: application/jsonrequired

Responses

Request samples

InClusterModelController_getAllModels

Responses

InClusterModelController_getModel

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required

Request Body schema: application/json
required