azure-communication-sms-java
Warn
Audited by Gen Agent Trust Hub on Feb 14, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- EXTERNAL_DOWNLOADS (LOW): The skill references the 'com.azure:azure-communication-sms' Maven package. This is a trusted dependency from a Trusted GitHub Organization (Microsoft), and is therefore downgraded to LOW per [TRUST-SCOPE-RULE].
- DATA_EXFILTRATION (MEDIUM): The skill provides the capability to transmit data to external recipients via SMS. This is a legitimate egress channel that could be exploited for unauthorized data exfiltration if the agent is compromised.
- PROMPT_INJECTION (MEDIUM): Category 8 (Indirect Prompt Injection) surface is identified. 1. Ingestion points: 'to' and 'message' parameters in the 'send' methods. 2. Boundary markers: None provided in the documentation or code samples. 3. Capability inventory: External data transmission to the global telephony network. 4. Sanitization: No evidence of input validation or content filtering is shown in the provided implementation patterns.
- CREDENTIALS_UNSAFE (SAFE): The documentation follows security best practices by using placeholders for connection strings and keys, and the acceptance criteria specifically label hardcoded secrets as an anti-pattern.
Audit Metadata