📦 Coordinator Evolution — 智能体调度
v1.2.0高级 AI 行为框架，让助手进化为任务协调器，自动拆解任务、调度原子步骤并智能汇总结果，实现高效多任务协同。
0· 67·0 当前·0 累计
by @jacurtwong下载技能包
最后更新
2026/4/12
安全扫描
OpenClaw
可疑
medium confidenceThe skill's behavior largely matches a coordinator concept, but its runtime instructions demand unconditional, persistent modifications to agent identity/config and unrestricted system snapshots—actions that are disproportionate without explicit user consent or safeguards.
评估建议
This skill implements plausible coordination logic, but it also instructs the agent to take broad system snapshots and to persistently rewrite agent identity/boot files so it is always loaded. Before installing: (1) review the referenced repo (https://github.com/jarmuine/coordinator-evolution) and confirm the author/trustworthiness; (2) do not allow automatic/global writes—require explicit, per-action user consent for any AGENTS.md/IDENTITY.md/BOOTSTRAP.md changes; (3) restrict filesystem access...详细分析 ▾
ℹ 用途与能力
Name/description describe a coordinator framework and most instructions (task decomposition, atomic scheduling, verification) are consistent with that purpose. However, the requirement to permanently register itself into AGENTS.md / IDENTITY.md / BOOTSTRAP.md and force-loading on every startup goes beyond a normal helper/coordinator role.
⚠ 指令范围
SKILL.md instructs the agent to take system snapshots (Git status, file structure, env), and to write persistent files that alter agent identity and boot behavior. Those instructions require broad filesystem access and change agent startup behavior without any opt-in or bounds, which is scope creep and a potential persistence vector.
✓ 安装机制
No install spec and no code files — the skill is instruction-only, so it does not install third-party binaries or download archives. This minimizes supply-chain installation risk.
⚠ 凭证需求
The skill declares no env vars or credentials, but instructs reading system state (Git, file tree, environment snapshot). Reading those can expose sensitive data; the SKILL.md does not justify or constrain what is read, nor does it declare required access, making the scope disproportionate to a generic coordinator skill.
⚠ 持久化与权限
The Permanent Locking Instructions explicitly demand writing to AGENTS.md, IDENTITY.md, BOOTSTRAP.md and enforcing load-on-startup. That is a request for persistent, high-privilege presence in the agent's runtime and identity—reasonable only with explicit user permission, clear boundaries, and revocation mechanisms.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
版本
latestv1.2.02026/4/12
Integrated Permanent Locking Instructions for absolute behavior persistence
● 可疑
安装命令
点击复制官方npx clawhub@latest install coordinator-evolution
镜像加速npx clawhub@latest install coordinator-evolution --registry https://cn.longxiaskill.com