The Agent Skills Directory

REMOTE_CODE_EXECUTION (CRITICAL): The installation instructions include a command that downloads a script from an untrusted remote source and pipes it directly into the shell (| sh). This is a highly dangerous pattern that bypasses package manager security checks and executes unverified code.
Evidence: SKILL.md contains curl --proto '=https' --tlsv1.2 -LsSf https://github.com/tursodatabase/turso/releases/latest/download/turso_cli-installer.sh | sh.
EXTERNAL_DOWNLOADS (MEDIUM): The skill instructs the user to install Node.js packages from an organization (@tursodatabase) that is not on the trusted sources list.
Evidence: npm i @tursodatabase/database in SKILL.md and references/javascript-api.md.
COMMAND_EXECUTION (LOW): The skill requires the execution of multiple local system commands for installation, database management, and encryption key generation.
Evidence: Extensive use of the tursodb CLI and openssl across all files.
PROMPT_INJECTION (LOW): The skill provides an interface to a database engine that processes external data and SQL queries, creating a surface for indirect prompt injection.
Ingestion points: Local database files (.db) and user-provided SQL queries.
Boundary markers: Absent.
Capability inventory: Local command execution (tursodb), file system write/read, and SQL execution.
Sanitization: No evidence of sanitization or instruction-ignoring delimiters.

turso-db