The Agent Skills Directory

Unverifiable Dependencies & Remote Code Execution (LOW): The skill installs Azure SDK crates using the Rust package manager.
Evidence: cargo add azure_messaging_eventhubs azure_identity and cargo add azure_messaging_eventhubs_checkpointstore_blob in SKILL.md.
Trust Status: These packages are maintained by the Azure GitHub organization, which is a verified trusted source. Under [TRUST-SCOPE-RULE], the download finding is downgraded to LOW.
Data Exposure & Exfiltration (INFO): The skill is designed to communicate with Azure cloud infrastructure over the network.
Evidence: Connection logic targets the <namespace>.servicebus.windows.net hostname.
Context: Network communication is the core functionality of the skill. It utilizes DeveloperToolsCredential for secure, non-hardcoded authentication, which is a best practice.
Indirect Prompt Injection (INFO): The skill ingests untrusted data from external Event Hub partitions.
Evidence Chain:
Ingestion points: receiver.receive_events(100, None) in SKILL.md.
Boundary markers: None present in the provided snippets.
Capability inventory: The skill allows for receiving and printing the body of external events.
Sanitization: Events are displayed via debug printing ({:?}) without specific sanitization.
Severity: INFO. In the context of this skill, the data is only used for local display (println!) and does not influence privileged agent decisions or command execution.

azure-eventhub-rust