The Agent Skills Directory

[SAFE]: The skill uses a bundled command-line tool (flowctl) to perform its core functions. This tool is executed from the skill's own internal directory, which is a secure and standard practice for bundled utilities.
[SAFE]: The skill processes user-provided feature requests to generate plans. It follows a 'no implementation code' policy, which serves as a security boundary by preventing the generation of executable code based on potentially untrusted input, ensuring the output is purely documentation and metadata.

flow-next-plan