← Back to Skills Marketplace
aaron-he-zhu

Memory Management

by Aaron Zhu · GitHub ↗ · v9.9.5 · MIT-0
cross-platform ✓ Security Clean
2122
Downloads
1
Stars
21
Active Installs
24
Versions
Install in OpenClaw
/install memory-management
Description
Persist SEO/GEO campaign context across Claude sessions with automatic hot-list, active work, and archive tiers. 项目记忆/跨会话
Usage Guidance
This skill appears internally consistent and implements a local file-based memory system. Before installing: 1) Verify the agent environment's memory/ folder permissions and backups so you don't lose or inadvertently overwrite important data; 2) If you plan to connect external SEO tools, confirm where those integrations are configured (this skill references them but does not declare credentials); 3) Test on a non-production project to observe how it promotes/demotes items and how audit blocks are appended; 4) If you have other skills that write the same wiki or audit files, decide which should be authoritative to avoid conflicts.
Capability Analysis
Type: OpenClaw Skill Name: memory-management Version: 9.9.5 The memory-management skill is a comprehensive state-persistence framework for SEO/GEO projects, utilizing a tiered memory system (HOT/WARM/COLD). It manages project context through structured markdown files in a dedicated 'memory/' directory and includes advanced features like a 'wiki layer' and GDPR-compliant data purging. While the instructions in SKILL.md are complex and direct the agent to perform file system operations (read, write, grep, delete), they are strictly scoped to project data and aligned with the stated purpose. Notably, the skill includes a defensive 'decision provenance' check to prevent the agent from treating its own inferred decisions as authoritative without user approval.
Capability Assessment
Purpose & Capability
Name/description (persist project memory across sessions) align with the declared behavior: the SKILL.md explicitly reads/writes memory/* files, manages promotion/demotion, builds a wiki index, and integrates with other skills. No unrelated credentials, binaries, or system-level config paths are requested.
Instruction Scope
Runtime instructions focus on creating/updating files under memory/ (hot-cache, audits, glossary, archives) and on lifecycle rules. The skill does reference conditional external tool actions ("If SEO tool connected: runs automated ranking check") but does not request or document any credentials or third-party endpoints. This is internally consistent but you should confirm how any connected SEO tools are configured elsewhere (other skills/integrations).
Install Mechanism
Instruction-only skill with no install spec and no code files — nothing will be downloaded or written by an installer. Lowest risk from install mechanism.
Credentials
No required environment variables, no primary credential, and no config paths declared. The operations (file reads/writes in memory/) are proportionate to a local memory-management skill.
Persistence & Privilege
always:false (normal). The skill claims ownership of wiki writes and will append audit blocks when explicitly instructed; that is within scope but means it will modify project files in the agent workspace. Confirm you are comfortable with a single skill being the authoritative writer for memory/wiki and memory/audits folders.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install memory-management
  3. After installation, invoke the skill by name or use /memory-management
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v9.9.5
Published from main @ 7ecc77b
v9.9.0
Published from main @ 7861a09
v9.5.0
Published from main @ 9967b3d
v9.1.0
Published from main @ d9bf8c7
v9.0.1
Published from main @ 3642433
v9.0.0
Published from main @ 032eb65
v8.0.2
Published from main @ 728f02d
v8.0.1
Published from main @ e644cbe
v8.0.0
v8.0.0: Unified version release — consolidates Wiki Knowledge Layer, Auditor Runbook inline strategy, Critical Fail Cap, Guardrail Negatives, and 2 new commands
v7.2.0
v7.2.0: conversion-focused first paragraphs, cross-promotion tables, deduplicated Chinese keywords, semantic density paragraphs for on-page/technical SEO auditors
v7.1.0
Improve ClawHub search discoverability: optimized display name, tags, and description for vector search
v7.0.0
v7.0.0: Wiki Knowledge Layer + infrastructure upgrades
v6.2.0
v6.2.0: when_to_use + argument-hint frontmatter, hook hardening, memory system upgrades
v6.0.0
v6.0.0: GStack pattern adoption — Completion Status Protocol, Escalation Protocol, Anti-Slop Output Voice, 8 named workflow phases, Decision Gates, AUTO-FIX vs ASK, 750+ multilingual triggers, all descriptions ≤180 UTF-8 bytes for full ClawHub display
v5.1.0
v5.1.0: multilingual trigger optimization — 5 languages, 750+ triggers, 8-agent reviewed
v5.0.0
v5.0.0: Unified operating model — hook automation, temperature memory, protocol gates, state write-through, trigger widening
v4.1.0
v4.1.0: publish GitHub absolute links for published docs and sync version metadata
v4.0.0
v4.0.0: ClawHub-first marketplace optimization — security fixes, vector search descriptions, multi-ecosystem install docs
v3.0.0
Version 3.0.0 of memory-management introduces major improvements and new reference resources. - Added 5 new reference files: examples, glossary-template, hot-cache-template, promotion-demotion-rules, and update-triggers-integration for better implementation guidance. - Expanded triggers and tags in metadata for improved discoverability and usage context. - Detailed compatibility information included for platforms and optional integrations. - Instructions now reference new template files for standardized memory structure. - Updated documentation to clarify context lookup flow and project glossary usage.
v2.0.0
No file changes detected in this version. - Version and documentation remain unchanged. - No new features, bug fixes, or updates introduced. - Functionality and instructions remain the same as the previous release.
Metadata
Slug memory-management
Version 9.9.5
License MIT-0
All-time Installs 23
Active Installs 21
Total Versions 24
Frequently Asked Questions

What is Memory Management?

Persist SEO/GEO campaign context across Claude sessions with automatic hot-list, active work, and archive tiers. 项目记忆/跨会话. It is an AI Agent Skill for Claude Code / OpenClaw, with 2122 downloads so far.

How do I install Memory Management?

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

Is Memory Management free?

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

Which platforms does Memory Management support?

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

Who created Memory Management?

It is built and maintained by Aaron Zhu (@aaron-he-zhu); the current version is v9.9.5.

💬 Comments