The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the presence of the mz-deploy command-line utility on the system path to provide LSP-based functionality such as object resolution and symbol search.
[PROMPT_INJECTION]: The skill ingests and analyzes content from .sql files in the user's workspace, which is a vector for indirect prompt injection where malicious code comments or identifiers could attempt to manipulate agent logic. 1. Ingestion points: Workspace SQL files in an mz-deploy project. 2. Boundary markers: None explicitly mentioned in the skill instructions. 3. Capability inventory: Uses the language server for hover, definition resolution, and symbol inspection. 4. Sanitization: Relies on the internal mz-deploy parser to validate project structure and syntax.

mz-sql-navigation