The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous commands to execute the glances monitor and perform administrative actions, such as running the tool with root privileges (sudo glances) or modifying system user groups for Docker socket access.
[EXTERNAL_DOWNLOADS]: Recommends installing the monitoring tool and its various plugin dependencies from well-known and trusted sources, including PyPI and official Linux distribution repositories.
[DATA_EXFILTRATION]: Documents features for exporting system health metrics to local CSV files or remote InfluxDB servers. It also details the built-in web server mode (glances -w) which exposes system performance data over the network on port 61208.
[PROMPT_INJECTION]: Identifies an indirect prompt injection surface where the agent processes system-generated data. 1. Ingestion points: The agent ingests glances output which includes untrusted host data like process names and container labels. 2. Boundary markers: The instructions provide no delimiters or safety warnings for the agent to ignore embedded instructions in the monitor output. 3. Capability inventory: The skill is used for executing system binaries and writing files (CSV export). 4. Sanitization: There is no defined process for sanitizing or escaping host data before the agent processes it.

glances