The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is designed to read and manage sensitive files within ~/.claude-marketing/credentials/ and ~/.claude-marketing/brands/. These directories house critical secrets including API keys, OAuth tokens, and environment variables. While essential for the skill's purpose, this represents significant exposure of high-value credentials.- [COMMAND_EXECUTION]: User-supplied input from the {slug} argument is interpolated directly into shell-like commands for local Python scripts (e.g., credential-manager.py --action get-profile --id {slug}). Without rigorous input sanitization, this provides a vector for arbitrary command execution on the host system.- [SAFE]: The skill's configuration includes disable-model-invocation: true, which is a prudent security measure that prevents the model from autonomously calling other tools or models while sensitive credential operations are being performed.

credential-switch