Writing Plans

Name/description (produce step-by-step implementation plans) align with the SKILL.md: it only asks the agent to create plan documents, list file paths, commands, and test steps. There are no unrelated environment variables, binaries, or install steps requested.

Instructions stay within planning: create plan files, enumerate exact file paths, provide code snippets, and include git/pytest commands and expected outputs. This is appropriate for an implementation-planning skill, but the SKILL.md explicitly requires embedding complete code and exact commands and references external sub-skills (superpowers:executing-plans, superpowers:subagent-driven-development). That increases the potential for subsequent automated execution (by other skills) — the plan itself does not perform I/O or network activity.

No install spec or downloaded code; instruction-only skills write nothing to disk by themselves. Lowest risk for install mechanism.

Requires no environment variables, credentials, or config paths. Nothing disproportionate requested relative to the stated purpose.

always:false and default autonomous invocation are set (normal). The main consideration is that the skill instructs use of other sub-skills that may execute plans, run tests, and commit code — evaluate those sub-skills' privileges before enabling autonomous invocation. The skill itself does not request persistent or system-wide privileges.