The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the ste-lint.py script using subprocess.run (in Python) and spawnSync (in Node.js) to evaluate text for style compliance. This execution is confined to the skill's own internal logic.
[PERSISTENCE]: The installation script install.py modifies the agent's global settings in ~/.claude/settings.json to register four hooks. This allows the skill's writing rules to persist across different chat sessions.
[PRIVILEGE_ESCALATION]: The skill performs setup actions that include creating symlinks and modifying the agent's configuration file. These actions are intended to enable the plugin's functionality within the host environment.
[INDIRECT_PROMPT_INJECTION]: The enforcement stack analyzes text generated by the agent or written to files, which constitutes a surface for indirect prompt injection. However, the risk is mitigated as the analysis is performed using deterministic regular expressions without executing the analyzed content.

asd-ste100