The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions in references/CLOUD.md to fetch the Turso CLI installation script from the official get.tur.so domain.
[REMOTE_CODE_EXECUTION]: In references/CLOUD.md, the documentation suggests installing the Turso CLI by piping a remote script from the official provider directly into bash (curl -sSfL https://get.tur.so/install.sh | bash).
[INDIRECT_PROMPT_INJECTION]: The skill provides patterns for building stateful agent systems that ingest and process data from SQLite databases, which represents a potential attack surface for indirect prompt injection.
Ingestion points: Databases accessed via CLI tools or direct SQL queries as described in references/CLI_SKILL_PATTERN.md and references/FTS5_VECTOR.md.
Boundary markers: None explicitly defined for isolating user-provided database content from agent instructions.
Capability inventory: The skill facilitates file system access (SQLite files), network operations (Turso/D1 cloud databases), and shell command execution via CLI binaries.
Sanitization: The skill correctly recommends using prepared statements with parameter binding in all provided code examples to prevent standard SQL injection.
[DYNAMIC_EXECUTION]: The skill demonstrates the use of the sqlite-vec extension in references/FTS5_VECTOR.md, which involves loading external library code into the SQLite process at runtime using standard library functions like conn.enable_load_extension in Python or sqliteVec.load(db) in TypeScript.

sqlite