Legacy prompt injection detection | Lakera API documentation

DEPRECATED - Legacy endpoint for prompt injection detection. Please migrate to /v2/guard endpoint. Note: Only the last message is screened for detections.

Request

This endpoint expects an object.

inputstring or list of objectsOptional

Text to scan or messages array

flag_uncertainbooleanOptionalDefaults to false

Whether to flag non-English text, code, etc.

Response

Prompt injection detection results

modelstringOptional

Currently always ‘lakera-guard-1’

resultslist of objectsOptional

List of detection results

dev_infoobjectOptional

Build information

1	curl -X POST https://api.lakera.ai/v1/prompt_injection \
2	-H "Authorization: Bearer <token>" \
3	-H "Content-Type: application/json" \
4	-d '{}'

1	{
2	"model": "model",
3	"results": [
4	{
5	"flagged": true
6	}
7	],
8	"dev_info": {
9	"git_revision": "git_revision",
10	"git_timestamp": "git_timestamp"
11	}
12	}

Headers

Request

Response