The Agent Skills Directory

[SAFE]: The skill functions as a read-only monitoring interface for the Olares platform. It utilizes the local olares-cli tool to gather system metrics such as CPU usage, memory allocation, disk health, and network performance.
[SAFE]: The instructions clearly define permission boundaries, noting that cross-user metrics require platform-admin authority, and emphasize the read-only nature of the tool.
[INDIRECT_PROMPT_INJECTION]: The skill processes system metric data returned by local command-line executions.
Ingestion points: System metrics and application resource rankings are ingested from the output of olares-cli dashboard subcommands as described in SKILL.md and references/olares-dashboard-overview.md.
Boundary markers: The skill uses structured JSON envelopes (Shape A and Shape B) to distinguish between metadata and raw machine data, reducing the risk of the agent misinterpreting data values as instructions.
Capability inventory: The skill is limited to read-only operations for monitoring and does not include capabilities for file writing, network exfiltration, or arbitrary command execution.
Sanitization: The reliance on structured JSON schemas for all metrics provides a validation layer for the ingested data.

olares-dashboard