The Agent Skills Directory

[SAFE]: The skill strictly follows a principle of least privilege, instructing the agent to utilize restricted SQL roles and to verify security settings using SHOW enable_rbac_checks before proceeding.- [SAFE]: Operational safety is built into the workflow by requiring that all tests and performance measurements be conducted on dedicated experiment clusters rather than live production environments.- [SAFE]: The instructions enforce an explicit authorization model, requiring the agent to obtain user permission for any data-reading operations such as cardinality probes or exactness proofs.- [SAFE]: The skill relies solely on official Materialize tools and system introspection views, with all external documentation links pointing to legitimate vendor-controlled domains.- [SAFE]: Thorough internal documentation within the development guide demonstrates a high level of security awareness, including the implementation of isolation protocols for the agent's execution environment.

mz-optimize-memory