The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes internal testing tools in the scripts directory that use the Python subprocess module to validate the orchestration graph integrity. These operations are limited to local validation of bundled JSON files and do not process untrusted external commands.
[INDIRECT_PROMPT_INJECTION]: As a routing component, the skill is designed to handle external text input. It mitigates potential injection risks by utilizing a strictly defined handoff contract and providing clear instructions to the agent to separate user data from execution intent.
[SAFE]: Architectural review shows robust design patterns including explicit ownership rules, bounded loops, and fallback mechanisms. The skill contains no hidden dependencies, obfuscation, or suspicious network activities.

avoid-ai-writing-router