Skip to content

Security: ApyGuard/ApyGuard-Benchmarks

Security

SECURITY.md

Security Policy

This repository intentionally includes vulnerable benchmark applications for security testing and evaluation.

Policy Scope

In-scenario vulnerabilities in benchmark apps are expected behavior and are not considered security defects for this repository.

Examples of intentional benchmark behavior:

  • broken access control
  • injection flaws
  • weak token handling
  • insecure object references
  • mass assignment

Please report issues that affect the repository, tooling, or supply chain instead.

Examples of in-scope reports:

  • accidental exposure of real credentials or private keys
  • vulnerable CI/CD or dependency-chain risks in project infrastructure
  • unsafe automation that can impact systems outside benchmark scope
  • repository permission or disclosure issues

Legal Disclaimer

These benchmarks are provided for educational and security research purposes only.

The maintainers are not responsible for any misuse, damage, or legal consequences resulting from:

  • deploying these applications in production environments
  • exposing them to public networks
  • using them for unauthorized testing against third-party systems

Users are solely responsible for complying with applicable laws and regulations.

Threat Model

These benchmarks are designed to simulate real-world API vulnerabilities including:

  • authorization bypass scenarios (BOLA, BFLA)
  • business logic abuse
  • token misuse and trust violations
  • data exposure patterns

They are intended to test:

  • API security scanners
  • behavior analysis systems
  • detection and triage workflows

Acceptable Use

You may:

  • run benchmarks locally or in isolated lab environments
  • use them to evaluate security tools
  • modify scenarios for research purposes

You must NOT:

  • deploy benchmarks to publicly accessible environments
  • use them to attack real systems
  • integrate them into production pipelines

Researcher Notes

If you are evaluating these benchmarks:

  • vulnerabilities are intentionally present and should be discoverable
  • focus on detection capability, not exploit novelty
  • some scenarios may require multi-step interaction flows

False positives or unclear behavior should be reported.

Out of Scope

The following are explicitly out of scope:

  • vulnerabilities inside benchmark scenarios
  • exploitation of intentionally vulnerable endpoints
  • denial-of-service attacks against benchmark apps
  • findings that require modifying benchmark assumptions

These are expected behaviors and will not be treated as valid reports.

Stability Notice

Benchmarks may change over time to improve realism or coverage.

  • vulnerability patterns may evolve
  • APIs and endpoints may be updated
  • backward compatibility is not guaranteed

Environment Isolation Requirements

Benchmarks must be executed in isolated environments:

  • containerized environments (Docker recommended)
  • no outbound internet access unless explicitly required
  • no access to internal or production systems

Failure to isolate environments may result in unintended security exposure.

Data Safety

All data used in benchmarks is synthetic.

  • no real user data is included
  • no production credentials are used
  • all tokens, keys, and flags are generated for testing purposes only

Benchmark Integrity

Do not patch or remove vulnerabilities unless explicitly required for testing.

These scenarios are designed to remain vulnerable by design.

How To Report

Use a private disclosure channel:

  1. Open a private vulnerability report using GitHub Security Advisories ("Report a vulnerability").
  2. If private reporting is unavailable, open a normal issue with minimal detail and request a private contact channel.

Do not post exploit details publicly before maintainers confirm triage.

Report Classification

Reports are evaluated based on impact:

  • High: Issues affecting repository security, infrastructure, or external systems
  • Medium: Issues in tooling or configurations with limited impact
  • Low: Documentation issues or minor inconsistencies

Reports related to intentional benchmark vulnerabilities will be closed as "by design".

Response Targets

Maintainers aim to:

  • acknowledge receipt within 3 business days
  • provide initial triage within 7 business days
  • share status updates at least every 14 days until resolution

Timelines are targets, not guarantees.

Coordinated Disclosure

Please allow maintainers reasonable time to investigate and remediate before public disclosure. When a fix is available, maintainers may publish an advisory and release notes.

Intended Audience

This repository is intended for:

  • security engineers
  • API security tool developers
  • researchers evaluating detection capabilities
  • organizations testing defensive systems

It is not intended for general-purpose application deployment.

Safe Use Requirements

  • Do not deploy these benchmarks in production
  • Do not expose them to the public internet
  • Run them only in isolated local or lab environments
  • Use synthetic credentials, flags, and data only

Security Contact

Security maintainer: Eren Kaan ÇAKIR

For sensitive disclosures, you may also contact:

Please avoid sharing sensitive details in public issues.

There aren't any published security advisories