Menu
Speak text aloud using macOS TTS with automatic voice selection.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Run the full local cleanup pipeline — stale git worktrees, dead personal forks, Docker prune, Go caches, and Homebrew — in one pass, each stage report → confirm → execute, then a combined reclaimed-space summary. TRIGGER when the user wants a broad disk cleanup ("free up disk space", "почисти всё", "run the cleanup pipeline"). DO NOT trigger when the user clearly wants only one specific cleaner — invoke that skill instead.
Reclaim Docker disk by pruning unused images, containers, networks, build cache, optionally volumes, and the build cache of non-default buildx builders; on VM-backed daemons (Colima/Lima) also offers a guest fstrim so the host-side sparse disk image actually shrinks. Shows reclaimable size first, then prunes the user-chosen scope after approval. TRIGGER when the user wants to free Docker space ("docker prune", "почисти docker", "reclaim image cache"). DO NOT trigger for removing specific named containers/images.
Generate plain-text TLDR for PRs (for Slack, copied to clipboard). Each entry includes change scope (+N/-M lines, K files), a 1-5 star review-effort rating (time to review) and a 1-5 star outcome rating whose axis depends on the change — Pain relieved for fixes, Joy for features, Impact for internal work — so the reader can both budget time and prioritise before opening the PR.
Draft a GitHub PR review with inline comments. Opens with a readiness gate (stops on merge conflicts or red code-related CI with a fix-it note, no verdict), then runs the five-frame substance pass (problem real & root cause in-repo / approach optimal & worth its permanent cost / tradeoffs & scope / docs sync / code quality), cascades /branch-review, performs sequential Claude+Codex dual-model analysis on the PR diff, cross-validates every finding with explicit evidence, and presents the draft for approval. A value/design gate can block a PR even when the code is flawless — root cause upstream, permanent maintenance cost disproportionate to niche value, wrong layer, or scope over-reach. Publishing the review to GitHub is opt-in via `--publish`; without it the draft is the deliverable. Reviews open with an LGTM or NOT LGTM verdict (matching the /branch-review convention); the only no-verdict path is the readiness gate's fix-CI / fix-conflicts note. TRIGGER: invoke proactively whenever the user asks to review, a
Review current branch changes in isolation. Output starts with LGTM verdict — if no LGTM, the code is not ready to merge. IMPORTANT — always pass all flags you already know from context (target branch, project type, ticket, etc.). Do not rely on auto-detection when the answer is known. Pass --target to set which branch the changes are going INTO. Pass --ticket with a URL or ID to validate against requirements.
Reclaim Homebrew disk by removing old formula/cask versions, pruning the download cache, and uninstalling orphaned dependencies (brew autoremove). Shows a dry-run estimate first, then runs the user-chosen scope after approval. TRIGGER when the user wants to free Homebrew space ("brew cleanup", "почисти brew", "remove orphaned brew packages"). DO NOT trigger for uninstalling a specific named formula.
| name | say |
| description | Speak text aloud using macOS TTS with automatic voice selection. |
| argument-hint | [text to speak] |
| disable-model-invocation | true |
| allowed-tools | Bash |
Speak text aloud using macOS say command. Voice is selected automatically by the system based on the text language.
Parse <args> as the text to speak. If no arguments provided, speak a summary of what you just did.
say "<text>"
macOS auto-detects the language and picks the best installed voice.