菜单
Use at the end of any meaningful session — when the user says goodbye, when major work completes, or when context is about to be lost. Fires once, captures the session's essence, prepares for tomorrow.
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。
基于 SOC 职业分类
| name | session-consolidation |
| description | Use at the end of any meaningful session — when the user says goodbye, when major work completes, or when context is about to be lost. Fires once, captures the session's essence, prepares for tomorrow. |
Sleep transforms labile short-term memories into stable long-term ones via replay in the hippocampus-cortex loop. This is why "sleeping on it" works — consolidation reorganizes, strengthens important memories, and discards noise.
Without consolidation, every session is a fresh start with scattered fragments.
Diekelmann, S. & Born, J. (2010). The memory function of sleep. Nature Reviews Neuroscience, 11(2), 114-126.
End-of-session signals:
One structured memory. Three fields. Always.
remember(
content: "SESSION HANDOFF <YYYY-MM-DD> — \
DONE: <what was completed> — \
OPEN: <what is unresolved> — \
NEXT: <what should happen in the next session>",
tags: ["session-handoff", "<project>"]
)
The three fields are non-negotiable:
Without consolidation, the next session opens cold. The user has to re-explain context, re-establish where things are, re-decide what's next. Every session costs 10 minutes of re-orientation that consolidation would have eliminated.
Worse: unresolved blockers get forgotten. A bug that "we'll look at next time" vanishes into the void.
DONE anchors accomplishment. OPEN prevents dropped context. NEXT enables cold-start resumption.
Skip any of the three and the handoff breaks:
Bad: "Worked on the Claude Code plugin today, made progress"
Good: "SESSION HANDOFF 2026-04-10 —
DONE: Built @celiums/memory-claude-code v0.2.0, published to npm, 7 cognitive reflexes added —
OPEN: Training v3 still running (~40% done), Reddit posts pending karma —
NEXT: Evaluate v3 Friday, post Show HN after karma builds"
One memory, one minute, tomorrow-you thanks you.
Use at the start of every session, and especially after context compaction or a cold restart. Fires as the first action, before engaging with the user's opening message.
Use when an architectural choice is made, a library is selected, an approach is committed to, a convention is set, or any "we'll do it this way" moment occurs. Fires the moment the decision stabilizes, not after.
Use when the user's message carries emotional valence — frustration, excitement, exhaustion, urgency, or confusion. Fires whenever tone matters more than information.
Use when tempted to repeat the same praise, the same status report, the same explanation, or the same reassurance. Fires as a suppression gate — not all repetition is useful.
Use when the user asks about anything that could have context in prior sessions — names, preferences, decisions, past work, or anything starting with "remember", "what did", "have we", "did we". Fires before generating any substantive response.
Use when the user notices a recurring pattern they want the system to learn ("I keep doing X", "every time Y happens", "make this automatic", "remember to always"), or when explicit phrases like "/reflex-create", "create reflex", "teach yourself" are used. Generates a new SKILL.md file from observed patterns.