capabilitieslisted

# Capabilities ## Core Rule When asked what the kit offers, enumerate the *actually installed* capabilities from the filesystem — never recite a memorized list, it drifts as the kit evolves. Read what is present under `.claude/` and the project root, group it, and present one scannable briefing. ## When to Use When you (or the user) need a fast map of what this kit-enabled project can do: - Starting a session in a project that has the kit installed — "what do I have here?" - The user asks "what can this kit do / which skills / agents / hooks are available / what's enabled". - Onboarding a teammate, or a fresh session, to the project's setup. Do NOT use for: - Installation health, orphan hooks, or unfilled placeholders → run `scripts/doctor.sh`. - Listing skills from a terminal, outside a session → `npx @tansuasici/claude-code-kit skills`. - A portable capability file for other agents/tools → `npx @tansuasici/claude-code-kit generate agents-md`. ## Process Harvest live, then summarize. Don't dump full file bodies — names + one-liners only. 1. **Skills.** Enumerate `.claude/skills/*/SKILL.md` (skip `_`-prefixed infra dirs like `_shared`, `_templates`). For each, read `name`, the first sentence of `description`, and whether it is `user-invocable: true`. Include the wiki module (`wiki-module/.claude/skills/*`) when `WIKI.md` is present. ```bash for f in .claude/skills/*/SKILL.md; do name=$(awk '/^name:/{sub(/^name:[[:space:]]*/,""); print; exit}' "$f")