← 返回 Skills 市场
jasoncodespace

X Engagement

作者 Jason · GitHub ↗ · v4.1.3 · MIT-0
cross-platform ✓ 安全检测通过
496
总下载
1
收藏
3
当前安装
5
版本数
在 OpenClaw 中安装
/install x-engagement
功能描述
X/Twitter 运营辅助。完整 onboarding → Persona 学习 → Browser Relay 浏览器控制 → 记忆系统 → 手动提醒模板 → For You 关注建议 → Following 互动建议
安全使用建议
This package appears coherent for its stated purpose, but before running anything: 1) Vet the Browser Relay tool (https://github.com/jasonCodeSpace/browser-relay) because the docs rely on npx/browser-relay‑cli which will download/execute third‑party code; only proceed if you trust that project. 2) Inspect the repo and scripts locally (especially any calls to npx or raw CDP commands) and confirm you are comfortable with a tool that can control your logged‑in browser. 3) Backup any existing memory/data you care about—cleanup scripts will delete files under ${HOME}/memory/daily/hotspots when run with --apply. 4) Keep the confirmation steps enabled and avoid automating the confirmation step; the skill is designed to require user confirmation before likes/follows/comments. 5) If you want to run this in a production account, test first on a throwaway account and review Browser Relay's source and npm package provenance.
功能分析
Type: OpenClaw Skill Name: x-engagement Version: 4.1.3 The x-engagement skill bundle is a well-documented Twitter/X automation assistant designed with significant safety guardrails. It utilizes an external 'Browser Relay' for browser interaction but explicitly mandates user confirmation for all 'write' actions (likes, follows, comments) and avoids creating persistent system tasks like crontabs, as seen in scripts/setup-cron.sh and docs/cron-jobs.md. The bundle includes a memory management system with safety checks in scripts/cleanup-memory.sh to prevent unauthorized file deletion, and the instructions for the AI agent (SKILL.md) prioritize human-in-the-loop verification and adherence to a specific persona without any indicators of data exfiltration or malicious intent.
能力标签
crypto
能力评估
Purpose & Capability
The name/description (X engagement / account operations) match what the skill actually does: onboarding, persona learning, generating candidate comments, and using a local browser runtime to execute confirmed write actions. It does not request unrelated credentials or hidden privileged access.
Instruction Scope
SKILL.md and docs instruct the agent to scrape tweets, generate candidate comments, check local history, and only perform likes/follows/comments after explicit user confirmation. That scope aligns with the stated purpose. Two notes: (1) the instructions rely on running npx browser-relay-cli commands (CDP/raw evaluation) to read DOM and interact with the logged‑in browser; (2) the skill writes and reads local files under ~/memory/daily/hotspots and the repo tree—this is expected but worth being aware of.
Install Mechanism
There is no formal install spec (instruction‑only), which lowers risk. However the runtime docs use npx browser-relay-cli and point to a third‑party GitHub repo; npx may fetch/run code from the npm registry on demand. The skill itself does not automatically download or run remote code, but following the docs will invoke third‑party tooling that the user should vet.
Credentials
The skill requests no environment variables or secrets. It operates on local browser sessions via Browser Relay and uses the user's HOME path for memory (memory/daily/hotspots). Those are proportionate to a social‑media automation assistant. No unrelated credentials or config paths are requested.
Persistence & Privilege
always:false and there is no attempt to modify system crontab by default; scripts explicitly avoid installing persistent schedulers. The skill writes to its own memory subtree and playbooks in the repo—expected for a learning/recording assistant. Note: the agent platform default allows autonomous invocation; this skill's docs emphasize manual confirmation for write actions, but if the agent were misconfigured to bypass confirmations the tool could act on the user's logged‑in browser (this is an operational risk, not an incoherence).
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install x-engagement
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /x-engagement 触发
  4. 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
v4.1.3
x-engagement v4.1.3 - 新增 _meta.json 元数据文件。 - 新增 runtime/manual-reminders.txt 文件,用于手动提醒配置或说明。 - No changes to existing source behavior. - 提升本地运行与手动提醒的文档和结构支持。
v4.1.2
清理数据文件:移除用户刷推记录,确保不包含任何个人信息或敏感数据
v4.1.1
新增评论规则:评论内容必须与推文相关(包含真实案例和操作流程)
v4.1.0
feat: 添加自我进化系统 - 新增 Playbook 系统(评论策略、变更记录) - 新增浏览器操作模块(基于 Chirp) - 新增评论规则(三个绝对禁止) - 新增数据采集功能 - 新增每日复盘脚本(22:00 自动运行) - 更新 SKILL.md(自我进化系统说明) 自我进化闭环:采集数据 → 分析对比 → 得出结论 → 更新规则 → 下次执行
v0.1.0
x-engagement v4.0.0 introduces major architectural and feature upgrades for automated X/Twitter operations. - 全新结构化设计,系统文件与文档分明,便于维护和扩展 - 增加三层记忆系统,支持评论历史、用户信息与热点事件记录 - 强化 Onboarding 流程,实现5阶段引导及 persona 学习 - 新增人类行为模拟,涵盖滚动、延迟、频率等多层仿真 - 自动化定时任务,支持每日热点总结与刷推提醒 - 评论逻辑优化,生成内容前自动避免自相矛盾
元数据
Slug x-engagement
版本 4.1.3
许可证 MIT-0
累计安装 3
当前安装数 3
历史版本数 5
常见问题

X Engagement 是什么?

X/Twitter 运营辅助。完整 onboarding → Persona 学习 → Browser Relay 浏览器控制 → 记忆系统 → 手动提醒模板 → For You 关注建议 → Following 互动建议. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 496 次。

如何安装 X Engagement?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install x-engagement」即可一键安装,无需额外配置。

X Engagement 是免费的吗?

是的,X Engagement 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

X Engagement 支持哪些平台?

X Engagement 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 X Engagement?

由 Jason(@jasoncodespace)开发并维护,当前版本 v4.1.3。

💬 留言讨论