Conciseness

The ~15-line body spends exactly one line on identity ("`gh stack` is a GitHub CLI extension for stacked branches and pull requests") and every remaining token on routing — no padding, no explanation of concepts Claude already knows. It matches the anchor "Lean and efficient; assumes Claude's competence; every token earns its place" rather than the 4 anchor, which reserves room for trimmable over-explanation that is absent here.

Actionability

The routing directives are concrete and unambiguous — named files with explicit trigger conditions ("before creating a stack or deciding layers", "on unexpected failures or for preconditions, side effects, and ordering guarantees") — so Claude knows exactly what to do next, and per the scoring note the absence of code in an instruction/routing skill is not penalized per se. It falls short of the 5 anchor ("copy-paste ready code or commands") because the body contains no inline command or quick-start example; any concrete action requires first opening a reference file.

Workflow Clarity

The body presents a clear navigation sequence (read usage.md first, then the trigger-matched reference), but the skill manages destructive/batch operations (merge, rebase, sync across a whole stack) and the body itself contains no validation or verification checkpoint — the cap in the judging guidelines therefore limits this to 3. It is not a 2 because the sequence that is present is coherent and each reference's purpose is well defined, and the references the body routes to do carry the validation loops (exit-code table, `gh stack view --json` to confirm, abort paths).

Progressive Disclosure

The body is a pure overview routing to four one-level-deep references, all of which exist in the bundle (references/usage.md, stack-design.md, commands.md, troubleshooting.md), each with a clearly signaled trigger; the reference files do not nest further (usage.md's pointer to stack-design.md is a sibling, not a chain). This matches the 5 anchor — clear overview, well-signaled one-level-deep references, easy navigation — with no organization gaps to justify a 4.

Specificity

The description lists nine concrete operations — "creation, viewing, editing, push, submit, sync, rebase, merge, and checkout" — which comprehensively covers gh-stack's command surface, matching the anchor for multiple specific concrete actions with comprehensive coverage. It is not a 4 because there are no meaningful gaps in the action list; it is not below 4 because every action is concrete rather than generic.

Completeness

It explicitly answers both questions: the "what" is managing stacked PRs via the enumerated operations, and the "when" is an explicit "Use when..." clause with multiple concrete trigger conditions. This mirrors the score-5 anchor example (concrete action list followed by explicit trigger phrases), and clearly exceeds the 4 anchor where the "when" is less specific.

Trigger Term Quality

Natural trigger phrases include "a stack", "branch layers", "gh stack", "stacked PRs", "dependent PRs", "splitting work into reviewable branches", and "a stack is already checked out" — the phrases a user would actually say, including synonyms. It sits above the 4 anchor because coverage extends beyond tool jargon to colloquial variations and a state-based trigger (stack already checked out).

Distinctiveness Conflict Risk

The niche is unambiguous — stacked PR management with the named tool gh-stack — and the triggers ("stack", "branch layers", "gh stack", "a stack is already checked out") are terms no other plausible skill would claim, giving minimal conflict risk. It does not fall to 4 because there is no meaningful overlap with neighboring git/GitHub skills' trigger terms.