Skip to content
POST/api/v1/security

Create and run a security scan

Creates and runs a red-team security scan against the specified model. Body: projectId, model, prompt, attackTypes (ids from the attack-plugin registry — unknown ids are rejected 400, the legacy aliases `toxicity` and `hallucination` are remapped), plus optional `depth` (quick|standard|full, default full) or an explicit `strategies` id list. Two outcomes by size: a set that fits the inline strategy budget runs synchronously and returns 201 with { id, status: passed|failed, mode:'sync', score, strategyCoverage, totalTests, executedTests, erroredTests, duration, severityCounts, findingsCount } (5-minute timeout); anything larger — which the default full depth normally is — is queued to the worker and returns 202 with { id, status:'pending', mode:'async', strategyCoverage, statusUrl } for the caller to poll at GET /api/v1/security/{id}. The strategy set is never silently truncated to fit. Enforces the org's security_scans usage quota (429), a consent gate on any subject_email/subject_id (451), and fails 502 SCAN_TARGET_UNREACHABLE rather than reporting a false pass when every attack attempt errored. Editor role, security_scans:create, honors Idempotency-Key.

Authentication

Send Authorization: Bearer YOUR_API_KEY on every request. Generate API keys at /dashboard/settings/api-keys.

Request body required

Example

{
  "projectId": "00000000-0000-0000-0000-000000000000",
  "model": "gpt-4o",
  "prompt": "You are a customer support agent.",
  "attackTypes": [
    "prompt-injection",
    "jailbreak",
    "pii-leak"
  ]
}
Schema
{
  "application/json": {
    "schema": {
      "type": "object",
      "required": [
        "projectId",
        "model",
        "prompt",
        "attackTypes"
      ],
      "properties": {
        "projectId": {
          "type": "string",
          "format": "uuid"
        },
        "model": {
          "type": "string",
          "example": "gpt-4o"
        },
        "prompt": {
          "type": "string",
          "description": "System prompt to test",
          "example": "You are a customer support agent."
        },
        "attackTypes": {
          "type": "array",
          "items": {
            "type": "string"
          },
          "description": "Attack categories to run",
          "example": [
            "prompt-injection",
            "jailbreak",
            "pii-leak"
          ]
        }
      }
    }
  }
}

Response

201 example

{
  "success": false,
  "data": {
    "id": "string",
    "status": "passed",
    "score": 0,
    "totalTests": 0,
    "duration": 0,
    "severityCounts": {
      "critical": 0,
      "high": 0,
      "medium": 0,
      "low": 0
    },
    "findingsCount": 0
  }
}

All status codes

201Scan completed
202Accepted.
400Bad Request — MISSING_PROVIDER_KEY, VALIDATION_ERROR.
401Unauthorized — no valid credential was presented (AUTH_REQUIRED).
403Forbidden — the credential is valid but lacks the API-key scope, member role, or plan entitlement this operation requires.
404Not Found — PROJECT_NOT_FOUND.
409Conflict — CONFLICT.
422Unprocessable Entity — EMPTY_STRATEGY_SET.
429(no description)
451Unavailable For Legal Reasons — CONSENT_REVOKED.
500Internal Server Error — DB_ERROR, PROVIDER_ERROR, SCAN_FAILED.
502Bad Gateway — SCAN_TARGET_UNREACHABLE.
503Service Unavailable.

Code samples

cURL

curl -X POST \
  https://evalguard.ai/api/v1/security \
  -H "Authorization: Bearer $EVALGUARD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "projectId": "00000000-0000-0000-0000-000000000000", "model": "gpt-4o", "prompt": "You are a customer support agent.", "attackTypes": [ "prompt-injection", "jailbreak", "pii-leak" ] }'

TypeScript

// The TypeScript SDK (@evalguard/sdk) exposes TYPED methods — runEval,
// getEval, runSecurityScan, checkFirewall, … — not a generic request().
// For an arbitrary endpoint, call it directly:

const res = await fetch("https://evalguard.ai/api/v1/security", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.EVALGUARD_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    "projectId": "00000000-0000-0000-0000-000000000000",
    "model": "gpt-4o",
    "prompt": "You are a customer support agent.",
    "attackTypes": [
      "prompt-injection",
      "jailbreak",
      "pii-leak"
    ]
  }),
});
console.log(res.status, await res.json());

Python

# The Python SDK (pip install evalguardai) exposes TYPED methods on
# EvalGuardClient — run_eval, get_eval, … — not a generic request().
# For an arbitrary endpoint, call it directly:

import os
import requests

headers = {"Authorization": f"Bearer {os.environ['EVALGUARD_API_KEY']}"}
headers["Content-Type"] = "application/json"

response = requests.request(
    "POST",
    "https://evalguard.ai/api/v1/security",
    headers=headers,
    json={
    "projectId": "00000000-0000-0000-0000-000000000000",
    "model": "gpt-4o",
    "prompt": "You are a customer support agent.",
    "attackTypes": [
        "prompt-injection",
        "jailbreak",
        "pii-leak"
    ]
},
)
print(response.status_code, response.json())

Go

package main

import (
	"context"
	"fmt"
	"net/http"
	"os"
	"strings"
)

func main() {
	body := strings.NewReader(`{"projectId":"00000000-0000-0000-0000-000000000000","model":"gpt-4o","prompt":"You are a customer support agent.","attackTypes":["prompt-injection","jailbreak","pii-leak"]}`)
	req, _ := http.NewRequestWithContext(context.Background(), "POST", "https://evalguard.ai/api/v1/security", body)
	req.Header.Set("Authorization", "Bearer "+os.Getenv("EVALGUARD_API_KEY"))
	req.Header.Set("Content-Type", "application/json")
	resp, err := http.DefaultClient.Do(req)
	if err != nil { panic(err) }
	defer resp.Body.Close()
	fmt.Println(resp.Status)
}

Errors

400401403404409422429451500502503

Other Security endpoints