The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill utilizes platform-specific dynamic execution syntax (!command) to run diagnostic probes (ops-speedup --scan) at load time. This is used to automatically populate the system health report and dashboard headers.
[COMMAND_EXECUTION]: The instructions direct the agent to execute a bundled binary (bin/ops-speedup) with various flags. These commands perform high-impact operations including process termination, Docker volume pruning, file deletion (caches, logs, temporary files), and network configuration changes.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external local files, including preferences.json and history.jsonl.
Ingestion points: cat and tail commands are used to read plugin data and telemetry logs for rendering timestamps and trend graphs.
Boundary markers: Not explicitly defined in the file-read instructions, though the data is parsed as JSON or TSV.
Capability inventory: The skill possesses extensive capabilities including shell command execution (Bash), file system access, and system-level configuration changes.
Sanitization: The skill relies on jq for parsing structured data from logs before display.
[PRIVILEGE_ESCALATION]: The skill performs operations that typically require elevated system privileges, such as kernel tuning (vnodes, somaxconn), flushing DNS caches, and modifying system startup items (launch-agents, systemd units). While sudo is not explicitly invoked in the instructions, these tasks represent high-privilege system modifications.

ops-speedup