← Back to Skills Marketplace
jdvrommel

Hermes Memory

by jdvrommel · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ Security Clean
114
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install hermes-memory
Description
Hermes风格自驱记忆系统。模仿Hermes Agent的记忆机制,定期自检记忆健康度、自动提醒整理、保持记忆精简高效。
README (SKILL.md)

Hermes风格自驱记忆系统

模仿Hermes Agent的记忆自检机制,定期检查记忆健康度并主动整理。

核心机制

文件 用途 建议大小
MEMORY.md Agent长期记忆 ≤2200字
USER.md 用户画像 ≤1375字
~/self-improving/ 技能记忆 按需

功能

  • 📊 记忆使用率监控
  • 💡 整理建议提醒
  • 🔄 自动存档旧记忆
  • ⏰ 定时自检(每天1次)

运行方式

手动自检:

python3 ~/self-improving/hermes-memory/memory_health.py

自动定时(每天收盘后17:30自检一次):

# 添加到crontab
crontab -e
# 写入:
30 17 * * 1-5 /home/linuxbrew/.linuxbrew/bin/python3 /home/jdvrommel/self-improving/hermes-memory/memory_health.py >> /home/jdvrommel/self-improving/hermes-memory/health.log 2>&1

触发整理的条件

  • MEMORY.md使用率 ≥70% → 建议精简
  • MEMORY.md使用率 ≥90% → 必须整理
  • USER.md使用率 ≥90% → 必须整理

和Hermes的区别

功能 Hermes 这个Skill
记忆限制 固定2200/1375字 建议上限
自动整理 Agent自动 提示+手动
记忆工具 add/replace/remove 文件编辑
快照机制 冻结快照 实时更新
Usage Guidance
This skill appears to do what it says: a local script that inspects MEMORY.md, USER.md, and files under ~/self-improving and writes a local state JSON. Before installing/running: (1) review the script yourself (it is short and included) to confirm you accept it reading the listed directories; (2) ensure the cron example paths (python interpreter and absolute home paths) are adapted to your system—do not copy the hardcoded /home/jdvrommel path; (3) be aware the tool will read potentially sensitive files in ~/.openclaw/workspace and ~/self-improving and will write hermes-memory-state.json there; (4) there is no network activity or credential access in the code, but if you change the script or integrate it with external tooling, re-check for exfiltration. If you want extra caution, run it manually once and inspect the generated state file before scheduling it in cron.
Capability Analysis
Type: OpenClaw Skill Name: hermes-memory Version: 1.0.0 The skill is a memory management utility for the OpenClaw agent, designed to monitor the size of memory-related markdown files (MEMORY.md, USER.md, etc.) and provide health reports. The Python script `memory_health.py` only performs local file read operations and basic character counting within the agent's workspace, with no evidence of data exfiltration, malicious execution, or obfuscation.
Capability Assessment
Purpose & Capability
Name/description (Hermes-style self-checking memory system) align with the included script and SKILL.md: the code inspects MEMORY.md and USER.md under the user's workspace and reports suggestions. The additional files referenced (SOUL.md, HEARTBEAT.md, ~/self-improving/) are plausible extensions of the described memory system.
Instruction Scope
SKILL.md and memory_health.py limit actions to reading memory files in the user's home (~/.openclaw/workspace/* and ~/self-improving/*) and writing a state file in ~/self-improving. This is within the stated purpose, but SKILL.md contains example cron lines with user-specific hardcoded paths (/home/jdvrommel and a linuxbrew python path) that don't match the script's use of Path.home(); users should update those before installing. The script reads all files in ~/self-improving which may include arbitrary sensitive content—this is expected for a memory-management tool but worth reviewing.
Install Mechanism
No install spec — the skill is instruction-only with one included Python script. Nothing is downloaded or executed from remote URLs, so there is no high-risk install behavior.
Credentials
No environment variables or external credentials are requested. The script accesses ~/.openclaw/workspace and ~/self-improving; that access is proportional to a local memory-management purpose, but those directories can contain sensitive data, so users should be aware the script will read them and will create a hermes-memory-state.json file in ~/self-improving.
Persistence & Privilege
The skill does not request elevated privileges, does not set always:true, and does not modify other skills or global agent settings. Cron-based periodic execution is suggested in SKILL.md but is optional and user-controlled.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install hermes-memory
  3. After installation, invoke the skill by name or use /hermes-memory
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
hermes-memory 1.0.0 发布,提供 Hermes 风格的自驱记忆自检和整理提醒工具。 - 支持定期(每天1次)自动检测 MEMORY.md 和 USER.md 文件的使用率 - 达到设定阈值时主动提醒进行记忆整理 - 自动存档和管理历史记忆,保持文件精简高效 - 提供手动运行和定时任务(cron)两种自检方式 - 与 Hermes Agent 的主要区别:整理建议为手动,记忆上限为建议而非硬性
Metadata
Slug hermes-memory
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Hermes Memory?

Hermes风格自驱记忆系统。模仿Hermes Agent的记忆机制,定期自检记忆健康度、自动提醒整理、保持记忆精简高效。 It is an AI Agent Skill for Claude Code / OpenClaw, with 114 downloads so far.

How do I install Hermes Memory?

Run "/install hermes-memory" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Hermes Memory free?

Yes, Hermes Memory is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Hermes Memory support?

Hermes Memory is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Hermes Memory?

It is built and maintained by jdvrommel (@jdvrommel); the current version is v1.0.0.

💬 Comments