跳到正文
TakoAPI
返回技能列表

opencortex

由 jd2005l 提供

Web & Frontend Development
939下载量由 clawskills.sh 提供
0浏览量
0 个赞登录后即可为这个技能点赞
—暂无评分
你的评分

简介

Self-improving memory architecture for OpenClaw agents.

描述

该技能的功能

Memory architecture system for OpenClaw agents. Structures memory into purpose-specific files and installs nightly distillation and weekly synthesis cron jobs that route daily work into permanent knowledge. Sensitive features (voice profiling, git backup, infrastructure collection) are off by default and require explicit opt-in.

Automates knowledge compounding through scheduled cron jobs rather than relying on manual memory management or flat context files.

何时使用

  • Setting up a new OpenClaw agent with organized memory from day one
  • Stopping an agent from re-asking questions or re-learning preferences it already encountered
  • Automatically backing up agent memory to git with secrets scrubbed before each push
  • Reviewing how much an agent has learned over time using compound growth scores
  • Recovering sub-agent work into the main memory pipeline via nightly distillation

示例工作流

输入

User asks: Set up memory architecture for my OpenClaw agent

智能体
  1. 1

    Runs clawhub install opencortex from the workspace root to download the skill

  2. 2

    Runs bash skills/opencortex/scripts/install.sh and selects optional features (vault, git backup)

  3. 3

    Installer creates the file hierarchy (SOUL.md, MEMORY.md, USER.md, memory/ subdirs) without overwriting existing files

  4. 4

    Cron jobs for nightly distillation (3 AM) and weekly synthesis (Sunday 5 AM) are registered

  5. 5

    Runs bash skills/opencortex/scripts/verify.sh to confirm files and cron jobs are active

输出

Structured memory files in place, principles enforced, nightly and weekly cron jobs running, agent begins compounding knowledge each session

前置要求

  • OpenClaw 2026.2.x+ installed
  • ClawHub CLI installed

安装

clawhub install jd2005l/opencortex

运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。

链接