memory-evolution - SKILL.md Agent Skill

name: memory-evolution description: Process pending X-OmniClaw task memories, update MEMORY.md, and rebuild the compact USER-PROFILE.md. metadata: { "xomniclaw": { "always": false, "emoji": "🧠", "skillKey": "memory-evolution", "version": "1.0.0", "category": "memory" }

}

Memory Evolution

Use this skill when the user asks to:

update global memory
evolve task memory
summarize task experience into the user profile
inspect memory evolution status
configure or verify the daily memory update mechanism

This skill maintains task-derived memories. Gallery-derived memories are maintained by gallery-memory.

Memory Files

MEMORY.md: task memory from X-OmniClaw usage, including habits, reusable workflows, failures, and project context.
memory/IMAGE-MEMORY.md: gallery memory from photos and screenshots.
memory/USER-PROFILE.md: compact user profile loaded by default. It should be rebuilt from both MEMORY.md and IMAGE-MEMORY.md.

Preferred Tool

`memory_evolution`

Use memory_evolution as the direct tool for this workflow.

Supported actions:

run: process pending task memory events, update MEMORY.md, and rebuild memory/USER-PROFILE.md
status: inspect the latest evolution run status and pending event count

Recommended Calls

Run evolution

{
  "action": "run"
}

Check status

{
  "action": "status"
}

Daily Update Pattern

The app owns a managed scheduled task named 全局记忆进化.

Its instruction should be equivalent to:

使用 memory_evolution 工具执行全局记忆进化：处理待沉淀的 X-OmniClaw 任务记忆，更新 MEMORY.md，并重建 USER-PROFILE.md。

Best Practices

Do not update MEMORY.md directly during every normal task. Normal tasks should only append pending events.
Keep USER-PROFILE.md compact because it is loaded by default for all tasks.
Put reusable task lessons and workflows in MEMORY.md; put only distilled user habits and key preferences in USER-PROFILE.md.
Do not duplicate gallery details in USER-PROFILE.md; summarize preferences, recent activity, and stable interests from IMAGE-MEMORY.md.