The Agent Skills Directory

[SAFE]: The skill is primarily instructional, guiding the agent on how to decompose requirements into a structured Markdown document. It does not perform any sensitive operations or access restricted data.
[COMMAND_EXECUTION]: The skill instructs the agent to include standard development commands (e.g., pytest, git commit) in the implementation plans it generates. These are contextual and appropriate for the intended software development use case.
[INDIRECT_PROMPT_INJECTION]: The skill acts on user-provided specifications. It includes a mitigation strategy in the form of a 'Plan Document Reviewer Prompt', which instructs a subagent to verify the plan against the original spec for alignment and completeness, reducing the risk of unintended behavior stemming from the input data.

writing-plans