From primitives
Cold-start from a handoff/plan document and execute the next step, then update the document so the next run can pick up where this one left off. Takes one argument: the @-tagged file. Built to be run repeatedly — /clear, /do-next @file, repeat. Use when the user says "do next", "do the next step", or hands you a handoff to continue.
How this skill is triggered — by the user, by Claude, or both
Slash command
/primitives:do-next @path/to/handoff-or-plan.md@path/to/handoff-or-plan.mdThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Read the @-tagged file as a cold agent, do the next step of work it describes, then leave the
Read the @-tagged file as a cold agent, do the next step of work it describes, then leave the document ready for the next run.
Read the file. It is the only context you have — treat it as the source of truth for what's done and what's next. Follow any artifacts it references (plans, ADRs, issues, commits) as needed.
Identify the next undone step and delegate it to /do-it — one step, the next vertical slice of
progress, not the whole backlog.
If the next step is ambiguous or blocked, stop and say so rather than guessing.
Invoke /update-handoff on the same file to fold this run's progress back in, so the next
/clear + /do-next cold-starts cleanly.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub xxkeefer/skills --plugin primitives