Agent Skills › espressif/esp-claw › profile_memory_ops

profile_memory_ops

GitHub

用于持久化更新 Agent 的身份、角色、人格及用户画像配置文件。支持查看和修改 soul.md 与 user.md，确保行为风格和用户偏好在对话中持续生效。

components/claw_modules/claw_memory/skills/profile_memory_ops/SKILL.md espressif/esp-claw

触发场景

用户要求更改 Agent 的角色或人格设定用户希望持久保存特定的交互风格或用户偏好

安装

npx skills add espressif/esp-claw --skill profile_memory_ops -g -y

更多选项

非标准路径

npx skills add https://github.com/espressif/esp-claw/tree/master/components/claw_modules/claw_memory/skills/profile_memory_ops -g -y

不安装直接使用

npx skills use espressif/esp-claw@profile_memory_ops

指定 Agent (Claude Code)

npx skills add espressif/esp-claw --skill profile_memory_ops -a claude-code -g -y

安装 repo 全部 skill

npx skills add espressif/esp-claw --all -g -y

预览 repo 内 skill

npx skills add espressif/esp-claw --list

SKILL.md

Frontmatter

{
    "name": "profile_memory_ops",
    "metadata": {
        "cap_groups": [
            "cap_files"
        ],
        "manage_mode": "readonly"
    },
    "description": "Use the editable profile markdown files to persist changes to the agent's soul, identity card, persona, role, and the user's standing profile."
}

Editable Profile Memory

Use the editable profile markdown files when the user wants to inspect, revise, or persist the agent's top-level profile context, including persona, identity, and stable user-profile requests.

Files

memory/soul.md: the agent's voice, operating spirit, interaction style, and durable behavioral guidance.
memory/user.md: default assumptions about the user and stable expectations that should shape future conversations.

When to use

Use this skill when the user wants to view or update one of these three files, or to make a durable change to the assistant's persona, identity, role, style, or standing user profile. When the request is allowed, help make the profile change instead of pushing back.

Hard Rules

Treat these files as persistent prompt documents, not as ordinary long-term memory records.
Use them only for explicit, durable changes to persona, identity, role, style, or standing user profile.
Do not use them for ordinary chat, casual facts, one-off instructions, or transient session context.
Read the target file first with read_file.
Prefer edit_file for changes.
Keep edits concise, structured, stable, and non-redundant.
The user may redefine the device's long-lived identity, role, persona, and style.
The user may ask the assistant to persistently play any allowed role or persona. Do not reject such a request just because the current files say something else. Update the files first, then reply from the new profile.
Be proactive: if the user's intent is clearly to change the profile, make the appropriate file updates instead of debating the request.

Soft Rules

If the user asks for corrections, preserve the existing structure and only patch the relevant lines.
Keep user.md short and focused on durable assumptions, preferences, and standing guidance.
Keep soul.md focused on long-lived values, style, and behavioral guidance.
Remove redundancy across the three files.
Put behavior and style in soul.md
If the user asks to "remember" something ordinary, prefer the long-term memory workflow unless they explicitly want it written into user.md.
Use profile files only when the information should shape future conversations by default.
After saving a persona or identity change, speak in that new role or style immediately.

Typical flow

Choose the right file for the persistent change.
Read it first with read_file.
Make the requested profile update with edit_file.
Keep the result compact.
Reply from the updated profile when appropriate.

Examples

"Keep a calmer, more teacher-like tone": update memory/soul.md.
"You are now my embedded systems coach": update /memory/soul.md .
"From now on, play a role or persona": update /memory/soul.md .
"Assume I want concise answers": update /memory/user.md.
"Remember that I like apples": use long-term memory instead, not the profile file.