Lean and directive throughout — no re-explanation of concepts Claude already knows, every section earning its tokens; the only mild redundancy (the Rule restating the trigger) reinforces rather than pads.

Provides concrete, executable commands such as 'gh issue view <number> --json state,url,title', 'rg -n "plan.md|tasks.md|docs/archives|NEEDS CLARIFICATION" docs AGENTS.md .agents/skills', and 'git status --short', plus specific keep/delete rules — copy-paste ready guidance for an instruction-only skill.

A clear five-step sequence is followed by an explicit Validation section with verification commands; because this destructive batch-delete workflow has concrete validation checkpoints, it is not capped at 2.

A simple single-purpose skill organized into well-labeled sections (Rule, Workflow, Cleanup Rules, GitHub Checks, Never Delete, Validation) with no need for external bundle references.

Lists multiple concrete actions — 'Scans docs/features, docs/issues, and docs/architecture', 'removes completed issue docs tied to closed GitHub issues', 'drops stale plan/tasks files', 'deletes obsolete feature or architecture docs' — matching the top anchor rather than the partial level 2.

Explicitly answers both: the 'Use only when a developer explicitly asks to clean, prune, tidy, or organize DeepChat SDD documentation after implementation and validation' clause supplies the trigger, and the rest states what it does.

Covers natural developer phrasing — 'clean, prune, tidy, or organize' plus 'SDD cleanup, documentation tidying, pruning, or removal' — giving good coverage of terms a user would actually say.

Narrowly scoped to DeepChat SDD documentation cleanup with distinct triggers, making it unlikely to fire for unrelated skills.