auvik-performance
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is designed for monitoring and observability, utilizing a local MCP server to interface with the Auvik API.
- [CREDENTIALS_UNSAFE]: The skill utilizes environment variables (AUVIK_USERNAME, AUVIK_API_KEY) for authentication via the Auvik API. This follows secure development practices for AI agents, and no hardcoded secrets or sensitive local files are accessed.
- [COMMAND_EXECUTION]: The skill invokes a local Python script (auvik_mcp_server.py) to handle MCP communications. The execution is scoped to the skill's defined functionality and does not allow for arbitrary command injection.
- [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests telemetry from an external monitoring platform. 1. Ingestion points: External data is ingested through tools querying Auvik performance endpoints (e.g., auvik_get_device_statistics, auvik_get_snmp_poller_history). 2. Boundary markers: The instructions do not specify explicit delimiters or warnings to the agent regarding potential injection in telemetry data. 3. Capability inventory: Capabilities are limited to executing the local auvik_mcp_server.py for read-only GET operations. 4. Sanitization: There is no mention of sanitization or validation of the numeric and string values returned by the API.
Audit Metadata