m365-entra-attack
Fail
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONCREDENTIALS_UNSAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill provides two comprehensive automation scripts:
- A Python script for ROPC (Resource Owner Password Credentials) validation that automates authentication requests to Microsoft's OAuth endpoints.
- A Playwright script for SAML SSO validation that performs browser-based automation to bypass or validate Conditional Access (CA) blocks.
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to install third-party security tools from public registries without integrity verification, specifically
msftreconando365sprayvia pip, as well as the ProjectDiscovery toolkit via Go. - [DATA_EXFILTRATION]: The provided automation scripts implement local logging of attack states, including email addresses, validation statuses, and attempt counters, to files like
engagement_log/o365_attempts.jsonando365_results.jsonl. - [CREDENTIALS_UNSAFE]: The Python validation script explicitly disables SSL/TLS certificate verification by setting
ssl.CERT_NONEandcheck_hostname=False. This critical security flaw exposes the agent to Man-in-the-Middle (MITM) attacks, allowing an attacker to intercept the credentials being sprayed. Additionally, the skill includes a list of common corporate password patterns to facilitate guessing attacks.
Recommendations
- AI detected serious security threats
Audit Metadata