Use when working in a knot-tracked project, signaled by `.knot.edn` or `.tickets/` at any ancestor of cwd, ids matching `<prefix>-01<base32>`, or intent like "what's next?", "what's blocked?", "list tickets", "show the backlog", "any pending bugs?", "what's open?", "what's tagged <x>?", "my tickets", "show me <id>", "track this", "open a ticket", "start <id>", "close this", "ship it", "add a note", or an autonomous agent told to pick up unblocked work. Do NOT use for hosted trackers (GitHub Issues, Linear, Jira, Basecamp, Asana, Trello) or for ids prefixed with hosted-tracker shortcodes (`GH-1234`, `ENG-1234`, `LIN-1234`, `JIRA-PROJ-1234`) — those have their own tools.
Token-efficient structural operations on Clojure/ClojureScript code via babashka + rewrite-clj. Use this skill BEFORE reading any .clj/.cljs/.cljc file over ~200 lines and BEFORE spawning Explore agents for Clojure code — measured 150× more token-efficient than reading whole files (single repo outlined in ~250ms, cross-repo grep across thousands of files in ~3s). Also trigger for: finding functions or usages across one or many repos, extracting or reordering defns, eliminating declares, renaming namespaces, converting CLJ↔CLJS↔CLJC, auditing dependency graphs of a single form. Triggers on phrases like "find where X is defined", "split this file", "what depends on Y", "extract these functions into a new ns", "is this declare needed", "rename this namespace", "convert these two files to cljc" — even when the user doesn't explicitly name the tool. Invoke as the `clj-surgeon` command on PATH (never `bb <script>`); every op is `clj-surgeon :op <command> :file <path> [:form <name> …]` with named args — run `clj-sur