The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using the sqlite3 CLI to interact with a local database file at /data/workspace/pantry.db. This is evident in the /fridge list, add, remove, and clear commands defined in SKILL.md.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted data and uses it in sensitive operations without sanitization.
Ingestion points: User input provided via the <item> and <quantity> arguments in the /fridge add and /fridge remove commands.
Boundary markers: None. User input is directly interpolated into the SQL command strings.
Capability inventory: The skill possesses command execution capabilities via shell calls to sqlite3 and file system modification (writing to the database file), as documented in SKILL.md.
Sanitization: No sanitization, escaping, or validation of user input is performed before interpolation into the SQL queries, creating a high risk of SQL injection.

fridge-tracker