← 返回 Skills 市场
Fluid Memory
作者
JiagHong Phang
· GitHub ↗
· v1.0.9
8046
总下载
5
收藏
38
当前安装
11
版本数
在 OpenClaw 中安装
/install fluid-memory
功能描述
基于艾宾浩斯遗忘曲线和访问频率的衰减模型设计的遗忘和归档机制,完全依赖openclaw原生记忆系统的拟人化流体记忆系统
安全使用建议
Install only if you want OpenClaw to keep a local, unencrypted memory of conversations. Enable the hook deliberately, avoid sending secrets while it is active, protect or periodically clear ~/.openclaw/workspace/database/, and do not run the daemon unless you want scheduled cleanup that can eventually delete archived memories.
能力评估
Purpose & Capability
The artifacts coherently implement local long-term memory, recall reinforcement, summarization, archiving, and forgetting behavior. These capabilities match the stated purpose, though they can retain personal or confidential chat content.
Instruction Scope
Tool triggers are memory-focused, but the optional hook records every sent message for later memory processing and the guidance encourages natural intent-based triggering. This is broad, but the hook behavior is documented in the hook and README.
Install Mechanism
Installation uses normal local skill files, Python dependencies, and an optional manual hook copy/configuration step. No hidden downloader, remote execution installer, or privilege escalation mechanism was found.
Credentials
The skill writes to ~/.openclaw/workspace/database/ and discloses plaintext local storage with no cloud sync. This is proportionate for a memory skill, but users should treat the database and logs as sensitive.
Persistence & Privilege
Conversation logs, ChromaDB memory, and summary buffers persist locally; a daemon can run scheduled maintenance if manually launched. Maintenance may permanently delete old archived memories, but it is limited to the skill's own memory store and no stealth startup was found.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install fluid-memory - 安装完成后,直接呼叫该 Skill 的名称或使用
/fluid-memory触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.9
feat: 更新描述+添加隐私声明
v1.0.8
fix: 修正HOOK.md描述+移除SKILL.md过时功能
v1.0.7
feat: 精简代码+重写简介,强调依赖原生flush触发
v1.0.6
feat: 5轮触发+Session恢复检查+完善互补说明
v1.0.5
feat: 优化静默模式 + 完善依赖声明
v1.0.4
feat: 动态查找Python路径,兼容不同环境
v1.0.3
feat: maintenance添加120天硬删除逻辑
v1.0.2
docs: 更新SKILL.md添加遗忘机制
v1.0.1
docs: 添加遗忘机制说明
v1.0.0
首发版本
v0.1.0
fluid-memory v0.1.0
- 初始发布:基于流体认知架构的记忆系统。
- 支持「植入记忆」「唤起记忆」「遗忘」和「状态查询」四大工具接口。
- 记忆会随时间/使用强化或遗忘,实现更自然、拟人化的记忆管理。
- 工具调用示例与最佳实践说明,便于理解与集成。
元数据
常见问题
Fluid Memory 是什么?
基于艾宾浩斯遗忘曲线和访问频率的衰减模型设计的遗忘和归档机制,完全依赖openclaw原生记忆系统的拟人化流体记忆系统. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 8046 次。
如何安装 Fluid Memory?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install fluid-memory」即可一键安装,无需额外配置。
Fluid Memory 是免费的吗?
是的,Fluid Memory 完全免费(开源免费),可自由下载、安装和使用。
Fluid Memory 支持哪些平台?
Fluid Memory 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 Fluid Memory?
由 JiagHong Phang(@againta)开发并维护,当前版本 v1.0.9。
推荐 Skills