ARES - authorized red-team engagement automation with dashboard, campaign scope, module orchestration, OPSEC controls, encrypted vault, and reporting.
Security Policy
Supported Versions
| Version | Supported |
|---|---|
| 6.x | Active security fixes |
| 5.x | End of life - upgrade to v6 |
| < 5.0 | End of life |
Reporting a Vulnerability
Please do not open a public GitHub issue for security vulnerabilities.
How to report
Email: [email protected]
Include in your report:
- Description of the vulnerability
- Steps to reproduce
- Potential impact assessment
- Any suggested mitigations (optional)
PGP: No project PGP key is currently published. Email first if encrypted coordination is required, and do not include sensitive exploit details until an encrypted channel has been agreed.
Response timeline
| Milestone | Target |
|---|---|
| Acknowledgment | 48 hours |
| Initial assessment | 5 business days |
| Fix development | 30 days for critical, 90 days for medium |
| Public disclosure | After fix is released + 7 days |
We follow coordinated vulnerability disclosure.
Scope
In scope
- Authentication bypass in the ARES API (
ares/api/) - Credential exposure - vault encryption, log redaction failures
- RBAC bypass - escalating from
reportertooperatororteam_lead - SQL injection or command injection in any module input handling
- Scope guard bypass - running modules against out-of-scope targets
- Dependency vulnerabilities with direct exploitability in ARES context
Out of scope
- Issues in modules that require compromising the target first (by design - ARES is a red team tool)
- Denial of service against the ARES API
- Issues in
[dev]optional dependencies - Social engineering
Security Design
API Response Data - Intentional Sensitive Output
EngineModuleResult.raw_output in API responses intentionally contains
captured hashes, Kerberos tickets, and credentials. This is the core value
of a red team tool - the operator needs this data to continue the engagement.
Protections applied to this data:
| Protection | Implementation |
|---|---|
| Authentication | require_operator() - only authenticated operators can run modules |
| Authorization | RBAC - reporter and recon roles cannot run credential modules |
| Transport | HTTPS enforced via nginx TLS + HSTS |
| Caching | Cache-Control: no-store on all API responses |
| Rate limiting | Per-endpoint limits prevent bulk extraction |
| Audit log | Every module_run_start and module_run_complete is logged with actor |
Operators should treat ARES API responses with the same sensitivity as the data they contain - store engagement results in encrypted storage.
Security Design
Key security controls in ARES v6:
| Control | Implementation |
|---|---|
| Credential encryption | AES-256-GCM (AEAD) with per-record PBKDF2-SHA256 salt (600,000 iterations) |
| Legacy encryption | Configurable via ARES_LEGACY_SALT env var |
| JWT tokens | HS256 default; RS256 supported via ARES_JWT_ALGORITHM=RS256 |
| Token revocation | JTI blacklist - logout immediately invalidates tokens |
| Scope enforcement | ScopeGuard hard-stops all modules; fails closed |
| Log redaction | Hashes, passwords, JWTs stripped from all structlog output |
| STEALTH enforcement | HIGH_NOISE modules raise ModuleValidationError before any network call |
| Subprocess isolation | Module crashes cannot affect the engine process |
| Input sanitization | sanitize_ldap() and sanitize_hostname() at 48+ call sites |
Known Security Considerations
This is a red team framework. It is designed to perform attacks against authorized targets. Deploying it in unauthorized environments is illegal.
Always:
- Obtain written authorization before running any ARES module
- Set scope via
ares campaign create --targets <cidr>before running modules - Use
--noise stealthfor sensitive engagements - Rotate
ARES_SECRET_KEYandARES_ENCRYPTION_KEYbetween engagements
This policy was last updated: 2026-03-21