The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill behaves as a standard cloud financial operations tool.
[COMMAND_EXECUTION]: The skill uses the finops CLI tool to analyze cloud data. This behavior is restricted to the specific commands required for cost analysis and resource discovery.
[DATA_EXFILTRATION]: The skill fetches cost data and resource metadata from AWS APIs. This data is stored in a local SQLite database at ~/.finops/data.db and is not shared with any unauthorized third-party services.
[PROMPT_INJECTION]: The skill ingests data from cloud resource tags and names, creating an indirect prompt injection surface.
Ingestion points: finops report resources provides external metadata to the agent context.
Boundary markers: Not explicitly specified in the CLI report commands.
Capability inventory: Bash(finops) capability allows execution of the local tool.
Sanitization: No specific sanitization logic is described for the ingested metadata. This surface is characteristic of resource discovery tools and does not pose a significant risk when handled by LLM guardrails.

finops