The Agent Skills Directory

[Prompt Injection] (SAFE): No instructions found that attempt to override agent behavior, bypass safety filters, or extract system prompts. The skill uses purely descriptive instructional language.
[Data Exposure & Exfiltration] (SAFE): No hardcoded credentials, sensitive file paths, or unauthorized network operations were detected. All external links point to trusted Microsoft documentation domains.
[Obfuscation] (SAFE): The content is clear-text markdown with no evidence of Base64 encoding, zero-width characters, or homoglyph-based evasion techniques.
[Unverifiable Dependencies & Remote Code Execution] (SAFE): The skill does not perform any external package installations or execute remote scripts. It references standard Azure MCP tools and CLI commands.
[Indirect Prompt Injection] (SAFE): The skill documentation identifies services that process external data, which is an inherent risk surface for indirect prompt injection, though no exploitable patterns are present in the skill itself.
Ingestion points: search_query results, speech_transcribe audio input, and Document Intelligence OCR data.
Boundary markers: None explicitly defined in this manifest.
Capability inventory: Communicates with Azure AI Service APIs for search and synthesis.
Sanitization: Relies on the host environment and specific tool implementations.
[Privilege Escalation] (SAFE): No commands for escalating system privileges or acquiring unauthorized permissions (e.g., sudo, chmod) are present.
[Persistence Mechanisms] (SAFE): No attempts to establish persistence, such as modifying shell profiles or crontabs, were detected.
[Metadata Poisoning] (SAFE): The metadata fields (name, description) accurately reflect the skill's purpose without deceptive or malicious content.
[Time-Delayed / Conditional Attacks] (SAFE): No logic gating behavior based on date, time, or environment variables was identified.
[Dynamic Execution] (SAFE): No runtime code generation, compilation, or dynamic library loading patterns were detected.

azure-ai