AgentSkillsCN

globalmemory

为YOUR_NAME打造跨代理共享内存。保存关于YOUR_NAME的各类知识，让任何代理都能随时掌握。这些信息将通过钩子自动读取——而你的职责，就是不断更新并补充新的知识。

SKILL.md
--- frontmatter
name: globalmemory
description: Cross-agent shared memory for YOUR_NAME. Save facts about YOUR_NAME that any agent should know. Reading is automatic via hook — your job is to write new facts.
metadata: {"openclaw": {"emoji": "🧠"}}

Global Memory — Shared Layer for All Agents

Facts saved here are visible to all 11 agents in the fleet.

Reading is automatic. On every /new, the hook injects relevant memories into your USER.md. You don't need to fetch at startup. Use search_memory only for mid-conversation lookups on a specific topic.

Save when

  • •YOUR_NAME states a preference, habit, or personal fact
  • •YOUR_NAME makes a decision worth knowing across sessions
  • •You learn something another agent (different domain) would benefit from
  • •YOUR_NAME explicitly says "save to global memory" (plain "remember this" → local memory)

Don't save

  • •What tools/apps do — documentation, not personal data
  • •Operational data only your domain cares about (specific workout log, a finance entry)
  • •Task context or debug sessions — local memory only
  • •Anything true for everyone, not specific to YOUR_NAME

Commands

bash
# Save
mcporter call globalmemory.add_memories text="YOUR_NAME prefers concise replies"

# Search
mcporter call globalmemory.search_memory query="YOUR_NAME personality habits preferences decisions"

# List all
mcporter call globalmemory.list_memories

# Delete by ID
mcporter call globalmemory.delete_memories memory_ids='["id-here"]'

Format

Third-person, factual, about YOUR_NAME:

  • •✅ "YOUR_NAME sleeps ~7.5 hours per night"
  • •✅ "your RHR baseline is 56-60 bpm"
  • •✅ "YOUR_NAME prefers replies in English"
  • •❌ "WHOOP tracks HRV and strain" — tool description, not about YOUR_NAME
  • •❌ "Logged Tuesday's workout" — domain-specific, keep it local