The Agent Skills Directory

[SAFE]: The skill serves as a procedural guide for automating reasoning paths and does not contain any instructions for data exfiltration, credential harvesting, or unauthorized system access.
[COMMAND_EXECUTION]: The instructions describe the use of an internal CLI utility lp (commands include lp score, lp registry add, and lp pending) used to manage the lifecycle of automated skills. These commands are integrated into the documented workflow for capability registration.
[DYNAMIC_EXECUTION]: While the skill governs the creation of new executable artifacts (scripts, SQL, workflows), it enforces strict safety protocols including independent acceptance gates, contract testing, and the exclusion of non-deterministic LLM calls from critical implementation logic.

crystallize-to-system1