cost-analysis

Pass

Audited by Gen Agent Trust Hub on Nov 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
  • Indirect Prompt Injection Surface: The skill processes external data from Azure environments, which constitutes a potential vulnerability surface.
  • Ingestion points: Data enters the context via query_costs, query_aks_costs, and Resource Graph query results as described in SKILL.md and referenced workflows.
  • Boundary markers: No explicit delimiters or 'ignore embedded instructions' warnings are provided for the data processed by the agent.
  • Capability inventory: The skill possesses read-only query capabilities and, with user approval, the ability to perform configuration changes (e.g., enabling AKS cost analysis or changing service tiers) via tools described in references/aks-enablement.md.
  • Sanitization: The risk is mitigated by requiring explicit user confirmation before any write operations or tier changes and emphasizing server-side aggregation.
  • Management Tool Integration: The skill uses established Azure management interfaces (MCP tools, Azure CLI, and PowerShell) to perform its functions.
  • This includes a fallback mechanism to command-line tools if specific MCP operations are unavailable, which is a standard pattern for administrative tasks.
  • The instructions in references/tools-and-safety.md emphasize using authenticated sessions and prohibit the exposure of access tokens or the use of local interpreters for data processing.
  • Operational Guardrails: Extensive guardrails are implemented to prevent unauthorized or accidental modifications.
  • The skill requires explicit approval for actions like upgrading cluster tiers, resizing node pools, or creating budgets.
  • It also includes instructions to report data limitations and maintain currency separation, ensuring the integrity and transparency of the financial analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 6, 2026, 08:20 AM