菜单
Ask the Second Brain to look now — runs one observer pass immediately over the focused task and reports each detector's verdict plus any advisory. Bypasses the pass throttle, not the mute or the budget. The fastest way to see it work.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/shofer-dev/shofer --skill second-brain-run命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | second-brain-run |
| description | Ask the Second Brain to look now — runs one observer pass immediately over the focused task and reports each detector's verdict plus any advisory. Bypasses the pass throttle, not the mute or the budget. The fastest way to see it work. |
Call the plugin request run (plugin second-brain, params { taskId? } — omit
taskId for the most recent observed task). It runs one pass immediately: every
enabled detector forks over the shared observation window and returns a verdict.
Relay each detector's verdict line verbatim (detector → silent | advise | resolved | still_open | timeout). An advise verdict may still be gated
(deduplicated, rate-limited, stale) — the note says so.
This bypasses the two limits that pace cost (the clock floor and the volume trigger) and neither of the ones that mean something: a mute still silences it, an exhausted budget still degrades to silence.
For a DEEP one-off investigation by a single detector, spawn its private mode as
a real task instead: new_task with mode second-brain:<detector> (e.g.
second-brain:git-log).