The Agent Skills Directory

[COMMAND_EXECUTION] (HIGH): The skill performs shell execution using node -e with unsanitized template variables.
Evidence: In SKILL.md, the 'Add Water Record' and 'Update Record' sections utilize a variable CUPS or NEW_CUPS directly inside a Node.js one-liner.
Risk: If a user provides input like 1); require('child_process').execSync('id'); //, the underlying command will execute that code instead of simply recording a number.
[REMOTE_CODE_EXECUTION] (HIGH): Vulnerable to Indirect Prompt Injection through unsafe interpolation.
Ingestion points: User input for the number of cups (CUPS, NEW_CUPS) is processed in SKILL.md.
Boundary markers: None. The placeholders are placed directly in the executable string.
Capability inventory: Full Node.js environment access, including file system (fs) and potential shell access via child_process.
Sanitization: Absent. There is no validation to ensure the input is a valid number before it is inserted into the script string.

healthcheck