The Agent Skills Directory

[DATA_EXFILTRATION]: The skill provides Python code snippets and instructions that access the sensitive Oracle Cloud Infrastructure (OCI) configuration file located at ~/.oci/config. This file typically contains API keys, fingerprints, and tenancy identifiers. Accessing this path is required for the intended functionality using the official oci Python SDK, but it constitutes an exposure of credential storage locations.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and displays infrastructure data which can be influenced by external environment factors, creating a potential surface for indirect prompt injection.
Ingestion points: Data is retrieved from the OCI Monitoring service via monitoring.summarize_metrics_data and monitoring.list_metrics in SKILL.md, specifically processing resourceDisplayName and metric values.
Boundary markers: The skill does not use specific delimiters or instructions to treat the OCI metric data as untrusted or to ignore potential commands embedded within resource names.
Capability inventory: The skill requires and uses the Bash(pip:*), Write, Edit, and Read tools to execute code and manage files.
Sanitization: There is no evidence of sanitization, escaping, or validation performed on the resource names or aggregated data points before they are output to the agent context.

oraclecloud-query-transform