The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill presents an indirect prompt injection surface because it processes untrusted SQL data and has file system access, although this is necessary for its stated purpose.
Ingestion points: The skill reads SQL queries from user-provided file paths or inline text messages in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the ingested SQL text.
Capability inventory: The skill is permitted to use write for saving translated SQL and altimate_core_validate for checking syntax, alongside file reading capabilities (SKILL.md).
Sanitization: No explicit validation or sanitization steps are defined for the input SQL before it is processed by the transpilation engine.

sql-translate