The Agent Skills Directory

[PERSISTENCE]: The skill implements a mechanism to establish automated triggers (afterFileEdit, beforeShellExecution) that persist within the developer environment. These hooks execute defined scripts every time a file is edited or a shell command is initiated, maintaining active execution across sessions.
[DYNAMIC_EXECUTION]: The skill generates new executable shell scripts (.sh files) at runtime and registers them for automatic execution by the platform. This dynamic generation and subsequent invocation of code allows for the creation of new behaviors not present in the original skill code.
[COMMAND_EXECUTION]: The generated scripts utilize the shell to run various commands. The skill provides examples for automated linting and testing, but the underlying capability permits the execution of any command-line operations whenever the specified event triggers occur.
[INDIRECT_PROMPT_INJECTION]: The skill identifies automation opportunities by processing user input and interaction history.
Ingestion points: User instructions and manual command patterns observed in the chat context (SKILL.md).
Boundary markers: Absent; there are no delimiters to isolate command patterns from potentially malicious instructions.
Capability inventory: Writing to the filesystem (.cursor/hooks.json and .sh files) and triggering automated shell execution.
Sanitization: Absent; the skill does not validate the safety of the commands being automated.

suggesting-cursor-hooks