The Agent Skills Directory

[NO_CODE]: The skill consists entirely of natural language instructions and configuration metadata. It does not include any executable scripts, binaries, or complex automation logic, which significantly reduces the potential attack surface.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill requests access to a local directory (such as a user's notes folder or an Obsidian vault) to manage food logs. This file system access is restricted to the primary functional purpose of the skill and is initiated by user setup. There are no network operations or instructions to exfiltrate this data to external services.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input, including meal photos and text descriptions, to extract nutritional data. While this is a common attack surface for indirect prompt injection, the instructions scope the agent's behavior to specific nutritional analysis tasks, minimizing the risk of arbitrary instruction execution from processed data.

meal-tracker