메뉴
Use when the user types /opencode or asks to run opencode for an external review, research pass, or second opinion. Runs the opencode CLI non-interactively in the background and reports back its verdict. Defaults to a strong reasoning model and supports alternates on request.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | opencode |
| description | Use when the user types /opencode or asks to run opencode for an external review, research pass, or second opinion. Runs the opencode CLI non-interactively in the background and reports back its verdict. Defaults to a strong reasoning model and supports alternates on request. |
Summon a fresh opencode CLI run as an independent outside model for a review, research pass, or second opinion. Use when the user writes /opencode "..." or asks to involve opencode.
opencode CLI installed (opencode --version).opencode/deepseek-v4-flash-free (the opencode Zen free tier) — no API key, just opencode auth login. Swap in any model you have configured — see Model selection.Run from the relevant workspace root:
opencode run "<prompt>" --model "opencode/deepseek-v4-flash-free" > /tmp/opencode-<tag>.log 2>&1
<tag> per call (a short topic slug).--dir <path> to root opencode elsewhere; --format json for machine-readable events.Run opencode models to list everything available to you, then pass --model "<provider>/<model>".
Free (opencode Zen — no API key):
opencode/deepseek-v4-flash-free — DeepSeek V4 Flash (the default)opencode/nemotron-3-ultra-free, opencode/mimo-v2.5-free, opencode/north-mini-code-freePremium (needs your own provider auth via opencode auth login) — stronger; the author's picks:
--model "deepseek/deepseek-v4-pro" --variant "max"--model "moonshot/kimi-k2.7-code" --variant "max"--model "qwen/qwen3.6-plus" --variant "max"--variant "max" only applies to providers that support variants — drop it for the free models.
The external model can take a while, so run it in the background and wait — don't block the turn:
run_in_background: true; you're re-invoked on exit, then read /tmp/opencode-<tag>.log.... &) and wait for exit, or run foreground with a generous timeout.Don't treat partial streamed output as final — wait for the process to exit, then summarize the concluding answer.
Include:
Check important findings against local code before recommending changes. Summarize the verdict, noting where you agree or disagree.
Use when the user types /claude or asks for a fresh Claude pass, an independent second opinion, or a clean-context review/research from Claude itself. Spawns a new headless Claude (`claude -p`) in the background — a separate context that isn't anchored to the current conversation — and reports back its verdict.
Generate images using Codex CLI's built-in image generation. Use when the user wants to create visual assets (sprites, icons, illustrations, mockups, game assets, UI elements) via the terminal. Wraps `codex exec` to delegate image generation to Codex, which has native access to the imagegen system skill.
Use when the user types /codex or asks for an external review, second opinion, sanity-check, or "ask codex" (RU «спроси codex», «внешнее мнение», «ревью от codex»). Also use proactively to pressure-test your own ideas, designs, plans, or diffs with an independent outside model before committing to them. Runs the codex CLI non-interactively in the background via `codex exec` and reports back its verdict.
Use when the user types /turbo-review or asks to convene multiple external models at once for a task, research, or review — a "quorum" pass. Fans the same prompt out to codex, opencode, and a fresh claude in parallel (read-only, in the background), then synthesizes their verdicts — flagging consensus (high confidence) and disagreement (needs scrutiny). Use proactively before committing to a risky design, plan, or diff.