auditing-entra-id-with-aadinternals

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The content and code explicitly document and automate creating federation backdoors (Golden SAML), forging SAML tokens, exporting token-signing certificates, and acquiring/caching access tokens—capabilities that enable account impersonation and persistent tenant compromise.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The runtime flow in scripts/agent.py performs unauthenticated outsider reconnaissance by fetching and parsing free-form JSON bodies from Microsoft endpoints (getuserrealm.srf and openid-configuration) derived from user-supplied domain, meaning outsider-authored text could be ingested and then acted on/printed.