AI Guard | Node.js SDK

AI Guard

constructor(token: string, config: PangeaConfig): AIGuardService

Creates a new AIGuardService with the given Pangea API token and configuration.

token

string

Pangea API token.

configPangeaConfig

Configuration.

AIGuardService

const config = new PangeaConfig({ domain: "pangea_domain" });
const aiGuard = new AIGuardService("pangea_token", config);

Text Guard for scanning LLM inputs and outputs

guardText(request: [object Object]): Promise<PangeaResponse<TextGuardResult<void>>>

Analyze and redact text to avoid manipulation of the model, addition of malicious content, and other undesirable data transfers.

request

Request parameters.

void

Promise<PangeaResponse<TextGuardResult<void>>>

const response = await aiGuard.guardText({
  text: "foobar",
});

Interface Detector

Detector

data

null | T

detected

boolean

Interface LogFields

LogFields

citations

string

Origin or source application of the event

extra_info

string

Stores supplementary details related to the event

model

string

Model used to perform the event

source

string

IP address of user or app or agent

tools

string

Tools used to perform the event

Interface MaliciousEntity

MaliciousEntity

action

string

type

string

value

string

rawRecord

start_pos

number

Interface PIIEntity

PIIEntity

action

string

type

string

value

string

start_pos

number

Interface TextGuardRequest

TextGuardRequest

debug

boolean

Setting this value to true will provide a detailed analysis of the text data

llm_info

string

Short string hint for the LLM Provider information

log_fieldsLogFields

Additional fields to include in activity log

recipe

[object Object] | pangea_prompt_guard

Recipe key of a configuration of data types and settings defined in the Pangea User Console. It specifies the rules that are to be applied to the text, such as defang malicious URLs.

Interface TextGuardResult

TextGuardResult

detectors

Result of the recipe analyzing and input prompt.

prompt_messagesT

Updated structured prompt, if applicable.

prompt_text

string

Updated prompt text, if applicable.