メニュー
Show stats/gain output from every companion tool that ships one, side by side. Use for /radin-stats, "show me all my stats", "aggregate my tool gains", "what savings am I getting from these tools".
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
Write a step-by-step implementation plan for one backlog entry, without touching code. Takes a task scope — a title/keyword — instead of the whole backlog. Judges whether the scope is broad enough to split into multiple independent plans, confirms any split with you directly, then writes one plan file per resulting sub-task, reviews it with thermo-nuclear and ponytail-review before handing it off, and appends a `**Plan:**` pointer back to the entry. Use for /radin-plan, "plan this backlog entry", "write a plan for X before we execute it". radin-execute also invokes this skill itself for any entry it judges too complex to implement without a plan.
Log feedback, bugs, follow-ups, or ideas raised mid-session as structured BACKLOG.md entries, so they survive past the conversation. Use for /radin-record, "log this to BACKLOG.md", "add as follow-up/bug/idea", "record what we just found", "note this for later", "add findings to backlog". Triggers even on vague asks ("add the findings") — scan the whole session, not just the literal text.
Run a thermo-nuclear code quality review over a scope (commit, PR, directory, or a range like "since yesterday") and log each finding as a BACKLOG.md entry instead of printing to terminal. Use for /radin-review, "review and log to backlog", "audit this commit/PR/directory and file backlog entries", "turn this review into a backlog".
Print the current project's BACKLOG.md to the terminal. Use for /radin-show, "show me the backlog", "what's in the backlog", "list backlog items", "print BACKLOG.md".
Run an extremely strict maintainability review for abstraction quality, giant files, and spaghetti-condition growth. Use for a thermo-nuclear code quality review, thermonuclear review, deep code quality audit, or especially harsh maintainability review.
Wire up per-repo hooks/MCP config for radin's companion tools (code-review-graph). Use for "set up hooks", "wire up code-review-graph", "enable the knowledge graph here", or right after install.sh in a new project.
SOC 職業分類に基づく
| name | radin-stats |
| description | Show stats/gain output from every companion tool that ships one, side by side. Use for /radin-stats, "show me all my stats", "aggregate my tool gains", "what savings am I getting from these tools". |
Surface each installed companion tool's own stats/gain command, back to back. No merged total: the numbers below use incompatible units (real per-session tokens, static benchmark medians, a counted per-repo ledger) — summing them would misrepresent all three, so display each as-is instead.
Run whichever of these are installed (command -v / skill lookup — skip
silently if missing, don't error):
/caveman-stats — real per-session token usage and savings, read from
the session log itself./ponytail-gain — ponytail's published benchmark scoreboard (medians
across 5 tasks, 3 models). Not this session, not this repo./ponytail-debt — ponytail's real per-repo ledger of deferred
shortcuts, if this repo has one.rtk gain — rtk's real token-savings ledger (command -v rtk; use
-p to scope to the current project).Any other installed tool with its own stats/gain command belongs here
too — add it to this list, don't build a separate skill.
If a source (e.g. caveman-stats) finds nothing to report, treat that as a
normal empty result — one quick check, then move on. Don't keep searching
for entries that aren't there. This is informative only, not exact/precise
required.
Print each tool's output under its own heading, in the order above. Label which are real-measured (caveman-stats, ponytail-debt, rtk gain) vs. fixed-benchmark (ponytail-gain) so the user doesn't mistake one for the other. Do not add a combined total row.
Each installed tool's own stats output, printed under its own heading, in the fixed order from Step 1. Each heading labeled real-measured or fixed-benchmark.