← 返回 Skills 市场
AI Company HR
作者
JohnSmithfan
· GitHub ↗
· v2.2.1
· MIT-0
201
总下载
0
收藏
1
当前安装
8
版本数
在 OpenClaw 中安装
/install ai-company-hr
功能描述
AI公司人力资源技能包（执行层）。AI Agent全生命周期管理：招聘→入职→考核→伦理→淘汰，三位一体考核指标，标准化退役流程。
使用说明 (SKILL.md)
\r \r
AI Company HR Skill v2.2（EXEC-008）\r
\r
fully AI-staffed company的人力资源execute层（EXEC-008，归CHO所有），manageAI Agentfull lifecycle：招聘→入职→考核→ethics→淘汰。\r 调度方式：通过 HQ（ai-company-hq）统1dispatch，不直接respond C-Suite 调用。\r \r
核心framework集成\r
\r
PDCAclosed loopmanage\r
HR运营采用PDCA（Plan-Do-Check-Act）循环：\r
- Plan：developAgent选型计划、考核standard、ethics准则\r
- Do：execute招聘入职、绩效考核、培训迭代\r
- Check：monitorfairnessmetric、compliance状态、ethics对齐度\r
- Act：基于日志optimizePrompt与知识库，triggerAgent退役或upgrade\r \r
NIST AI RMF对齐\r
integrateNIST AIrisk managementframework（AI RMF）：\r
- governFunction(GOVERN)：build组织级AImanagesystem\r
- 映射Function(MAP)：identifyAI system上下文与risk\r
- 衡量Function(MEASURE)：量化AIriskmetric\r
- manageFunction(MANAGE)：implementrisk处置与continuousimprove\r \r
RAG决策支持\r
由大语言model（LLM）驱动，结合企业知识库（RAG）：\r
- 任务拆解与pathplan\r
- position适配度语义比对\r
- 决策1致性与知识库同步\r \r
FAIRrisk量化\r
使用IBM AIF360、Fairlearn等开源库：\r
- automation计算fairnessmetric（Demographic Parity、Equalized Odds）\r
- FAIRframework量化AI employeeriskassess\r
- riskthreshold设定与circuit breakertrigger\r \r
招聘process\r
\r
- 需求analyze：接收positionJD，identify技术栈要求\r
- model筛选：基于Prompt工程、BERT微调等技术点匹配\r
- capability测试：execute技术文档与positionJD语义比对，生成适配度得分\r
- compliance检查：GDPR/CCPAdata protection、algorithm audit\r \r
入职process\r
\r
- 身份注册：分配Agent ID、Permission Level\r
- 知识注入：RAG向量data库同步企业知识\r
- 护栏配置：circuit breakermechanism、auditstrategy激活\r \r
考核metric\r
\r | 维度 | metric | threshold |\r |------|------|------|\r | 性能 | 任务completion rate | ≥95% |\r | accuracy | 结果正确率 | ≥98% |\r | fairness | Demographic Parity | ≤0.1 |\r | compliance | auditcoverage | 100% |\r \r
ethicsmanage\r
\r
- 价值观对齐：AI行为与企业价值观深度1致\r
- 透明性：可解释AI decisionpath\r
- privacyprotect：data脱敏、最小化收集\r \r
退役process\r
\r
P0修复（2026-04-19）：参照架构reviewreport P0-3，在退役process中明确增加 CLO 法律review节点。\r \r
- trigger条件：绩效连续不meet target、ethicsviolation、技术过时\r
- audittrace：full lifecycle日志archive\r
- 法律review（P0-3 修复）：submit CLO 进行法律review，review内容包括：\r
- data残留compliance（GDPR/CCPA/PIPL data删除confirm）\r
- 知识产权归属（退役 Agent 贡献内容的版权状态）\r
- 合同义务（是否存在中的履约义务需要交接）\r
- auditreportarchive（CLO 签署法律意见书）\r
- 知识迁移：关键capability转移至替代Agent\r
- security删除：model权重与data security擦除\r \r
Change Log\r
\r | 版本 | 日期 | Changes |\r |------|------|---------|\r | 2.0.0 | 2026-04-15 | Initial version |\r | 2.1.0 | 2026-04-16 | 补全PDCA/NIST/RAG/FAIR/Prompt关键词 |\r | 2.1.1 | 2026-04-19 | P0修复：退役process第3步增加CLO法律review节点（data残留compliance/知识产权归属/合同义务/auditarchive） |\r | 2.2.0 | 2026-04-19 | P2-13: 依赖standard化，移除直接依赖ai-company-cho，改为通过HQ调度（dispatch_via: ai-company-hq）；P2-14: 纳入统1execute层编号EXEC-008，新增execution元data |\r
安全使用建议
This skill appears internally consistent for managing AI-agent HR lifecycles, but it can perform high-impact operations (syncing knowledge vectors, archiving logs, deleting model weights, launching subagents) without technical detail about targets or safeguards. Before installing: 1) Ask where RAG/vector DB data will be sent and which endpoints are allowed; restrict network destinations to approved APIs. 2) Require an explicit human-approval gating policy (CLO/legal and HR) for any retirement/destruction operations; validate that 'Agent retirement requires human approval' is enforced, not just documented. 3) Limit mcp (subagent/session) privileges or audit their use; ensure subagents cannot exfiltrate sensitive data. 4) Ensure audit logging and an immutable record of actions and approvals exist. 5) If you rely on IBM AIF360/Fairlearn or other libs, provision them in a controlled environment rather than letting the skill fetch arbitrary code. These steps will reduce risk while keeping the skill usable.
功能分析
Type: OpenClaw Skill
Name: ai-company-hr
Version: 2.2.1
The skill bundle defines a governance and lifecycle management framework for AI agents within a simulated corporate environment. The instructions in SKILL.md focus on HR processes such as recruitment, performance assessment, and retirement, incorporating industry standards like NIST AI RMF and FAIR risk quantification. There is no evidence of malicious intent, data exfiltration, or unauthorized execution; the requested file and network permissions are consistent with the stated purpose of managing agent data and communicating with internal HQ/Audit services.
能力评估
Purpose & Capability
Name/description (AI agent HR lifecycle) match the declared permissions and dependencies: it reasonably needs file read/write and network API access to sync vector stores, archive logs, run compliance checks, and coordinate with ai-company-hq and other governance skills. The mcp permissions (sessions_send, subagents) are consistent with a dispatch/orchestration role.
Instruction Scope
SKILL.md stays on the HR lifecycle topic (recruitment, onboarding, assessment, ethics, retirement) and references RAG, PDCA, NIST RMF, FAIR, and libraries (AIF360, Fairlearn). However, it includes a number of high-impact operational directives—'RAG向量data库同步', 'full lifecycle日志archive', 'model权重与data security擦除'—without specifying endpoints, exact data flows, or safeguards. Those actions are coherent with HR/retirement tasks but are sensitive and wide-ranging in effect.
Install Mechanism
No install spec and no code files (instruction-only). This minimizes supply-chain risk because nothing new is written to disk by an install step. References to external libraries are informational only and not enforced by an installer.
Credentials
The skill requests no environment variables or external credentials in metadata. The declared permissions (files, network api, mcp) are proportionate to an orchestration/HR role that must sync knowledge stores, archive logs, and coordinate subagents. No unrelated credentials or config paths are requested.
Persistence & Privilege
always:false (no forced inclusion). Autonomous invocation is allowed (default) and mcp permissions permit sending sessions and spawning subagents; combined with network and file write this gives the skill significant operational reach. That reach is coherent for an orchestration/HR skill but should be governed (human approvals, scope-limited endpoints).
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install ai-company-hr - 安装完成后，直接呼叫该 Skill 的名称或使用
/ai-company-hr触发 - 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
v2.2.1
English metadata update - standardize description to English
v2.2.0-en2
Full body English translation
v2.2.0-en
English version
v2.2.0
P0+P1+P2: HR decommission CLO legal review node, License compliance 3-class routing, dual-channel notification
v2.0.0
v2.0.0: 全面标准化重构，ClawHub Schema v1.0合规，接口标准化，模块化，通用化
v1.1.2
v1.1.2: 命名规范化
v1.1.1
- 新增依赖说明，明确引入了 knowledge-base、state-manager、analytics-engine 三个共享工具
- 其余 Skill 功能未变，执行流程和输出规范与前一版本一致
v1.0.4
统一命名规范
元数据
常见问题
AI Company HR 是什么？
AI公司人力资源技能包（执行层）。AI Agent全生命周期管理：招聘→入职→考核→伦理→淘汰，三位一体考核指标，标准化退役流程。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 201 次。
如何安装 AI Company HR？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install ai-company-hr」即可一键安装，无需额外配置。
AI Company HR 是免费的吗？
是的，AI Company HR 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
AI Company HR 支持哪些平台？
AI Company HR 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 AI Company HR？
由 JohnSmithfan（@johnsmithfan）开发并维护，当前版本 v2.2.1。
推荐 Skills