The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a bash script (scripts/spawn-agent.sh) to manage agent sessions via the herdr and jq CLI tools. User inputs like name and task are passed as arguments to these tools. The script correctly uses double-quoting for variable interpolation to prevent shell word splitting. While this involves subprocess execution, it is standard for the skill's purpose and relies on the platform's underlying tool execution logic for security.
[INDIRECT_PROMPT_INJECTION]: The capability to spawn agents and send them tasks creates a surface for indirect prompt injection. However, the skill explicitly mitigates this through detailed 'Anti-Patterns' in SKILL.md. These instructions strictly forbid the agent from inferring tasks or session names from external data (e.g., ticket IDs) without explicit user confirmation, ensuring that sub-agents are only spawned and task-prompted upon direct request.
[SAFE]: No malicious patterns such as data exfiltration, credential harvesting, obfuscation, or persistence mechanisms were found. The skill follows best practices for tool development, including environment validation and automated cleanup of resources on failure.

spawn-agent