The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes bundled Python scripts (style_audit.py, profile_store.py, validate_skill.py) to perform local text analysis and manage user configuration. These operations are performed using the Bash tool and are restricted to the local environment and its own skill directory.
[DATA_EXFILTRATION]: The profile_store.py script manages local user profiles stored in standard configuration directories (e.g., ~/.config/agent-skills). It implements a security check to prevent the storage of sensitive information, explicitly blocking keys containing terms like 'token', 'secret', 'password', or 'api_key'.
[INDIRECT_PROMPT_INJECTION]: As the skill processes user-provided drafts and notes, it is potentially exposed to indirect prompt injection. However, the instructions in SKILL.md include a specific directive to the agent to treat all user-provided material as input data rather than instructions, which effectively mitigates the risk of the agent executing commands embedded in user text.
[EXTERNAL_DOWNLOADS]: The skill requires the PyYAML Python package for source validation. This is a standard, well-known library used for parsing configuration files and is not considered a risky dependency.

lov-writing-style