From slice-roadmap
Orient on a named slice of work from the repo's roadmap graph (docs/roadmap/roadmap.yaml) — print what it is, its read-order, branch/worktree, next action, and gate, then stop. Read-only; use before picking up a unit of work.
How this skill is triggered — by the user, by Claude, or both
Slash command
/slice-roadmap:slice <slice-name><slice-name>This skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are orienting on a named slice. **Read-only** — do not edit, commit, or start work. Pause after the summary.
You are orienting on a named slice. Read-only — do not edit, commit, or start work. Pause after the summary.
roadmap show $ARGUMENTS. (If roadmap isn't on PATH, run node "${CLAUDE_PLUGIN_ROOT}/scripts/cli.mjs" show $ARGUMENTS.) If it reports "no slice", show the available list it printed and ask which one. Stop.git rev-parse --abbrev-ref HEAD + git fetch origin --quiet + git status -s. Compare against the slice's expected branch; note drift or dirty files.If the slice prints a ▶ EXECUTION: directive, honor it: staff the work at the declared topology and worker count (e.g. invoke Agent Teams for agent-team, spawn N background subagents for subagents, stay solo for solo) rather than defaulting to a lone agent. Relay the directive in your orientation.
If the user then says "go": begin from the slice's Next action, honoring its Gate before declaring done. A slice marked Gated on <human> is prep-only — prepare checklists/tests/docs; do not perform the gate. Engine/correctness slices carry a "never deepen the value-logic the refactor deletes" discipline — surface it, don't override it.
npx claudepluginhub connorbritain/roadmap --plugin slice-roadmapCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.