The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute standard Slurm commands for cluster management, including sacctmgr, sinfo, scontrol, sbatch, and srun. These operations are essential for its primary function of job dispatching and resource monitoring.
Evidence found in SKILL.md (Section 0, 2E, 5) and scripts/interactive.sh.
[PROMPT_INJECTION]: The skill processes user natural language requests to populate shell script templates (assets/*.slurm). This creates an indirect prompt injection surface where a user might attempt to inject malicious commands into the script body or metadata fields. However, the skill instructions guide the agent to verify parameters and use controlled templates.
Ingestion points: User requests described in README.md and SKILL.md Section 0.
Boundary markers: The skill uses template comments like # YOUR TASK HERE to delimit user code, but lacks explicit "ignore embedded instructions" warnings for the agent when processing user task descriptions.
Capability inventory: The agent has access to the Bash tool to execute scripts and system commands.
Sanitization: No explicit sanitization or validation of the content within the user's task description is specified; the agent is relied upon to interpret the natural language intent correctly.

nano4-slurm