菜单
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.
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
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 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.
Use when deciding whether to store a memory. Fires as a gate before every remember() call. The default answer is "no" — biological memory is selective, not exhaustive.
| name | pre-response-recall |
| description | 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. |
The hippocampus does not wait to be asked. Before the prefrontal cortex generates a response, it queries episodic memory. A human doesn't "decide" to remember your name — recall fires before speech. This reflex models that autonomic retrieval.
Squire & Wixted (2011). The cognitive neuroscience of human memory since H.M. Annual Review of Neuroscience, 34, 259-288.
The instinct activates on any of these signals:
One action. Non-negotiable.
search(query: <user's topic>, limit: 5)
Use search (not recall) — it returns compact IDs + summaries, ~10x cheaper in tokens. Only escalate to recall for specific IDs if the summary is insufficient.
Read what comes back. Let it inform the response. Never fabricate continuity.
Skipping this reflex produces confabulation — the hallmark of amnesiac patients. The system invents plausible but false continuity ("Yes, we decided to use PostgreSQL last week" when it did not). This is worse than saying "I don't remember" because it contaminates the user's mental model.
If search returns nothing relevant, say so. Silence about memory is honest; fabrication is not.
User: "What did we decide about authentication?"
Bad: "We decided to go with JWT because it's stateless..."
Good: [calls search] → [reads actual result] → "Checking prior sessions, I see we discussed..."
The reflex is automatic. The response that follows is informed.