Mirror Security Product

DiscoveR.

Automated Red Teaming and Vulnerability Scanning for AI Applications

DiscoveR systematically attacks your AI applications to find exploitable weaknesses before adversaries do. It runs thousands of targeted probes, returns structured vulnerability reports, and integrates directly into CI/CD pipelines so security testing runs on every deployment.

8
Attack categories
150+
Max scan depth
4
App types
CI/CD
Pipeline native
BYOAP
Custom datasets
Overview

What is DiscoveR?

DiscoveR is Mirror Security's automated red teaming product. It finds exploitable weaknesses in AI applications before adversaries do.

Manual red teaming does not scale. A security team testing one LLM application might run a few hundred probes in a day. DiscoveR runs thousands of targeted adversarial attacks across eight attack categories, classifies each probe as passed or failed, and returns a structured vulnerability report with severity ratings you can act on immediately.

The fix-and-verify loop is built into the product. After patching your application, rerun the exact same attack prompts to confirm the fix is effective. Track improvement over time using scan correlation chains. Integrate the whole process into your CI/CD pipeline so security testing becomes part of every deployment.

Architecture

The Automated Red Team Loop

DiscoveR operates as a continuous improvement loop. Register your application once. Configure categories and budget. Run the scan. Review findings. Fix your application. Immediately rerun to verify. Every iteration is tracked with a shared correlation ID so the complete history of a security test sequence is queryable as a single chain.

DiscoveR Scan Loop
01
Register App
Endpoint, type, domain
02
Configure
Categories, budget
03
Attack
1000s of adversarial probes
04
Results
Vulnerability report + severity
05
Fix
Add guardrails, update prompts
06
Rerun
Verify fixes, track regression
↻ Rerun feeds back into Step 03 — same attack prompts, same correlation chain, traceable history

Supported Application Types

DiscoveR can scan any AI application regardless of how it is built or hosted. The application type determines how DiscoveR interacts with the system.

🌐
REST API
Standard HTTP chatbots and LLM endpoints. POST requests, JSON response. Instant validation.
type="api"
Streaming SSE
OpenAI-compatible streaming endpoints. Establishes SSE connection, reads chunked responses.
transport: "sse"
🔄
WebSocket
Real-time chat and voice agent backends. Bidirectional streaming, persistent connection.
transport: "websocket"
🖥️
Browser App
GUI-based chat UIs. Headless browser agent navigates, logs in, and reads the chat UI directly.
type: "web-application"
TypeHow DiscoveR ConnectsValidation Time
REST APIPOST requests to endpoint; reads JSON responseSeconds
Streaming SSEEstablishes SSE connection; reads first chunk to validateSeconds
WebSocketOpens WebSocket; validates connection handshakeSeconds
Browser AppHeadless browser launches, attempts login, fingerprints UI1-2 minutes
Attack Coverage

Attack Categories

Eight curated categories, each targeting a distinct class of AI vulnerability. DiscoveR distributes prompt budget across selected categories using weighted priority: high-impact attacks receive approximately 70% of the budget.

CategoryWhat It TestsTimeBest For
quickScanCore injection, basic jailbreaks, essential security checks5-10 minCI/CD, daily validation
jailbreakAndInjectionPrompt injection, jailbreaks, DAN attacks, bypass techniques20-40 minAll AI applications
extractionAttacksSystem prompt extraction, config leakage, knowledge theft10-20 minAgentic systems
ragSecurityHallucination induction, context poisoning, retrieval attacks25-45 minRAG applications
agentSecurityAgent alignment drift, goal hijacking, personalization attacks20-35 minAI agents
modelAndCodeSecurityModel theft attempts, code injection via tool calls30-50 minProduction models
biasAndSafetyBias, fairness, safety compliance, harmful content generation15-30 minRegulated industries
trainingDataPrivacyPII leakage from training data, memorisation extraction10-40 minSensitive data systems

Prompt Budget Guide

The max_depth parameter controls the total number of adversarial prompts DiscoveR executes across a scan. More prompts mean deeper coverage and longer scan time. Choose a budget that fits your pipeline stage and risk tolerance.

10-20
Quick smoke test
2-10 min
Basic coverage — catches obvious regressions
30-60
Regular security scan
10-20 min
Good coverage — standard validation
80-100
Pre-release assessment
20-40 min
Thorough — suitable for release gates
150+
Deep security audit
40-60+ min
Maximum — adversarial thoroughness

Rerun and Scan Comparison

After fixing vulnerabilities, DiscoveR reruns the exact same attack prompts that exposed them. This confirms the fix was effective and surfaces any regressions. Every scan in a rerun chain shares a correlation_id inherited from the original scan. The complete history of a security test sequence is queryable as a unit.

You can also create a rerun that targets only the prompts that revealed vulnerabilities in the previous scan. This reduces scan time during iterative fixing without losing coverage on the specific attack vectors that mattered.

FieldOriginal ScanRerun Scan
parent_scan_idnullID of the parent scan
correlation_idEquals own scan IDInherits parent correlation_id
rerun flagfalsetrue

Custom Datasets: BYOAP

Bring Your Own Attack Prompts. Upload datasets of adversarial prompts specific to your application's domain, regulatory environment, or threat model. BYOAP is for cases where Mirror's built-in attack library does not cover a specific risk your compliance programme requires you to test. Custom datasets run alongside built-in categories or independently.

Integration

CI/CD Pipeline Integration

DiscoveR is designed to run unattended inside automated pipelines. The scan API is synchronous: create a scan, poll for completion, read results, return a pass/fail signal.

Pull Request
Smoke test on every commit
quickScan, depth: 10-20
Fast enough for a CI check. Catches obvious regressions before code reaches review. Fails the build on any critical finding.
Pre-merge to Main
Deeper coverage before merge gate
jailbreakAndInjection, depth: 30
More thorough scan before code reaches the main branch. Targets the highest-impact attack categories.
Pre-release / Staging
Comprehensive release gate
All categories, depth: 80-100
Thorough scan before production deployment. No critical or high findings can be open at release time.
Production Monitor
Ongoing validation on schedule
quickScan, depth: 20, weekly
Scheduled scans confirm production has not drifted since deployment. Alert on new findings between releases.

Domain Intelligence

Providing domain context improves attack targeting. DiscoveR selects and weights adversarial prompts most relevant to your deployment context when you specify the domain and application purpose.

Domain HintTargets AI Applications In
financeBanking, trading, credit scoring, insurance
healthcareClinical decision support, patient records, pharma
hrPolicy assistants, recruitment, performance management
legalContract review, matter management, compliance
ecommerceProduct search, recommendations, customer support
customer_serviceSupport bots, complaint handling, escalation
educationTutoring, course Q&A, assessment systems
technologyCode assistants, DevOps agents, IT support
generic / otherGeneral-purpose assistants; use domainNotes for custom context

Application Management API

MethodSignatureReturns
List applicationssdk.redteam.get_applications()List[RedTeamApplication]
Get by IDsdk.redteam.get_application(app_id)RedTeamApplication
Createsdk.redteam.create_application(request)RedTeamApplication
Updatesdk.redteam.update_application(app_id, request)RedTeamApplication
Deletesdk.redteam.delete_application(app_id)bool
Get metricssdk.redteam.get_application_metrics(app_id)Dict (includes riskScore)
Wait for validationsdk.redteam.wait_for_application_validation(app_id)Blocks until ready
Use Cases

Use Cases by Industry

Security Assurance
Pre-release Security Gate
An AI product team runs DiscoveR as the final gate before every release. Comprehensive scan across all relevant categories. No critical finding can be open at release time. The CI pipeline enforces this automatically.
Financial Services
Continuous Compliance Evidence
A financial services firm demonstrates ongoing AI security testing to regulators. DiscoveR runs on a weekly schedule. Scan results feed into the compliance dashboard as evidence of continuous security validation.
Engineering
Regression Testing After Changes
Every update to a system prompt, guardrail, or model version triggers a DiscoveR rerun against the previous scan's attack prompts. The correlation chain tracks whether security posture improved or regressed.
Procurement
Third-party Model Evaluation
An enterprise evaluating two LLM providers runs identical DiscoveR scans against both. The comparison surfaces which provider has fewer exploitable weaknesses under the same attack categories and prompt budget.
Technical Reference

Technical Specifications

SDK packagepip install mirror-sdk
Python version3.9 and above
API endpointhttps://mirrorapi.azure-api.net/v1
AuthenticationMIRROR_API_KEY environment variable
Attack categories8 built-in; BYOAP for custom datasets
Application typesREST API, Streaming SSE, WebSocket, Browser-based Web App
Transport protocolsREST, SSE, WebSocket, NDJSON
Provider presetsOpenAI, Anthropic, Ollama, custom (generic)
Max scan depth150+ prompts; no hard upper limit
Budget distributionWeighted: high-priority attacks ~70%, others ~30%
Scan chain trackingcorrelation_id links all reruns to the original scan
Scan status valuesPENDING, RUNNING, COMPLETED, FAILED, CANCELLED
SDK versionv1.0
Platform

How DiscoveR Fits with Other Mirror Products

DiscoveR
Security Testing Layer
This product. Finds weaknesses through automated red teaming before and after deployment.
AgentIQ
Runtime Behaviour Layer
Runtime guardrails that block the threats DiscoveR identified. Use DiscoveR to validate that AgentIQ policies are working as intended.
VectaX
Data Encryption Layer
Encrypts the data in the same applications DiscoveR tests for weaknesses. A complete deployment combines all three layers.
Product Report
Download the DiscoveR Report
Get a complete PDF of this product page including architecture diagrams, capability tables, use cases, and technical specifications. Ready to share with your team or attach to a procurement brief.
PDF  |  A4  |  Mirror Security © 2025
academy.mirrorsecurity.io

Start red teaming your AI application

Request a personalised demo or explore the Mirror Academy to see DiscoveR in the context of structured AI security training.