메뉴
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.
기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.
설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
직접 명령은 검토 Prompt를 거치지 않습니다. 실행하기 전에 소스를 확인하세요.
npx skills add https://github.com/shofer-dev/shofer --skill second-brain-run명령은 한 줄로 유지됩니다. 복사하기 전에 가로로 스크롤해 전체 내용을 확인하세요.
로컬 사본을 원하시나요? SkillsMP에서 현재 제공할 수 있는 파일을 다운로드하세요.
View or change the Live Memory plugin's configuration for this workspace — the provider profile the memory LLM uses, retention caps, the external-edit watch glob, the compaction interval, and the context-window budget/threshold. Use when the user asks to point Live Memory at a different model/provider, tune how much it remembers, or change its context budget.
Provides context about the Shofer evals system structure in this monorepo. Use when tasks mention "evals", "evaluation", "eval runs", "eval exercises", or working with the evals infrastructure. Helps distinguish between the evals execution system (packages/evals, apps/web-evals) and the public website evals display page (apps/web-shofer-code/src/app/evals).
Show what the Second Brain has observed, spent and said — observed volume per segment, pass latency, window fill, token use and cost, advisories generated / gated / delivered, and uptake per detector. Use when the user asks whether the Second Brain is watching, what it has cost, or whether its advice is being taken.
SOC 직업 분류 기준
SKILL.md 표시 중
| 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).