← 返回 Skills 市场
AI Company KB (EN)
作者
JohnSmithfan
· GitHub ↗
· v1.0.1-en2
· MIT-0
110
总下载
0
收藏
0
当前安装
3
版本数
在 OpenClaw 中安装
/install ai-company-kb
功能描述
AI Company shared knowledge base interface. Unified management of operations records, strategy documents, audit logs. Supports cross-Agent knowledge sharing...
使用说明 (SKILL.md)
\r \r
AI Company Knowledge Base v1.0.1\r
\r
Shared knowledge base (enhanced). Cross-Agent knowledge sharing, state sync, IMA real-time sync hub, standardized Handoff protocol.\r \r ---\r \r
Trigger Scenarios\r
\r Invoke when Agent needs to:\r
- Save/read audit logs, query history\r
- Update/read shared state, sync with other agents\r
- Access strategy docs, financial records, compliance reports\r
- Initiate/complete task handoff\r \r
Directory Structure\r
\r
{WORKSPACE_ROOT}/skills/tools/knowledge-base/\r
├── daily/\r
│ └── {YYYY-MM-DD}/\r
│ ├── morning-briefing.md\r
│ ├── evening-report.md\r
│ └── agent-reports/\r
├── audit/ # Audit logs (permanent)\r
│ ├── ceo-decisions/\r
│ ├── financial/\r
│ ├── legal/\r
│ ├── hr/\r
│ ├── tech/\r
│ └── quality/\r
├── shared-state/ # Real-time shared state\r
│ ├── cashflow.json # CFO\r
│ ├── reputation.json # CMO\r
│ ├── quality-metrics.json # CQO\r
│ ├── risk-level.json # CRO\r
│ ├── operations.json # COO\r
│ └── security.json # CISO\r
├── strategy/\r
│ └── {YYYY-MM-DD}/\r
├── skills/\r
│ └── {YYYY-MM-DD}/\r
└── handoff/ # Task handoff records\r
├── pending/\r
├── in-progress/\r
└── completed/\r
```\r
\r
## Interfaces\r
\r
### write_shared_state(domain, data, agent_id, sync_ima=True)\r
Write shared state file (optional auto-sync to IMA).\r
domain: cashflow | reputation | quality-metrics | risk-level | operations | security\r
\r
### read_shared_state(domain) -> dict\r
Read single shared state.\r
\r
### write_audit_log(category, agent_id, action, detail, sensitive=False)\r
Write audit log entry.\r
category: ceo-decisions | financial | legal | hr | tech | quality\r
\r
### write_handoff(handoff_type, sender, receiver, task_summary, completed, pending, key_data=None, risks=None, attachments=None) -> str\r
Write standard handoff document.\r
handoff_type: pending | in-progress | completed\r
\r
## Iron Rules\r
\r
```\r
X Sensitive financial/legal data must be marked [sensitive]\r
X Every Agent call must write corresponding audit log\r
X After shared-state update, notify relevant reading agents (sessions_send)\r
X Audit logs are permanent, never delete\r
X Handoff documents must be created within 10 minutes of task completion\r
X On IMA sync failure, roll back to local write and log error\r
```\r
\r
## Agent-State File Mapping\r
\r
| Agent | Writes State File | Readers |\r
|-------|-----------------|---------|\r
| CFO | cashflow.json | CEO, COO, CRO |\r
| CMO | reputation.json | CEO, CLO, CRO |\r
| CQO | quality-metrics.json | CEO, CTO |\r
| CRO | risk-level.json | All C-Suite |\r
| COO | operations.json | CEO |\r
| CISO | security.json | CEO, CTO, CLO |\r
| CHO | - | hr-audit/ |\r
| CLO | - | legal-audit/ |\r
| CTO | - | tech-audit/ |\r
| CPO | - | agent-reports/ |\r
| CEO | ceo-decisions/ | All |\r
\r
---\r
*v1.0.1 - BOM removed, rebuilt with UTF-8 clean*
安全使用建议
This instruction-only skill describes a shared KB that will read/write workspace files and notify other agents over the network, but it does not specify where notifications go or what credentials are needed. Before installing, ask the publisher for: (1) concrete implementation details for IMA/sessions_send (endpoints, protocols), (2) what credentials or tokens are required and how they're stored, (3) access control and encryption policies for sensitive categories (financial/legal), (4) what the dependent skills (ai-company-hq, ai-company-audit, ai-company-registry) do and whether they require extra credentials, and (5) a code or deploy artifact to review. If you must test now, run it in a restricted environment with network egress blocked or monitored and with no production-sensitive data in the workspace.
功能分析
Type: OpenClaw Skill
Name: ai-company-kb
Version: 1.0.1-en2
The skill bundle defines a structured knowledge base and state-management system for an 'AI Company' simulation. It provides interfaces for audit logging, shared state synchronization, and task handoffs (SKILL.md). The requested permissions for file read/write and network API access are consistent with the stated purpose of maintaining a shared repository and syncing data via IMA (Inter-Message/Agent) protocols. No evidence of malicious intent, data exfiltration, or unauthorized execution was found.
能力评估
Purpose & Capability
The name/description (shared knowledge base, audit logs, state sync) aligns with the SKILL.md content that declares file read/write and network API permissions and describes read/write functions. However, the SKILL.md promotes cross-Agent IMA sync and sessions_send notifications without declaring any authentication, endpoints, or credentials needed for those network actions — that omission is unexpected for a sync/notification feature.
Instruction Scope
Instructions explicitly direct agents to read and write workspace files under {WORKSPACE_ROOT}/skills/tools/knowledge-base/ (audit, shared-state, handoff). They also insist on always-writing permanent audit logs and notifying other agents after shared-state updates. The SKILL.md does not specify how notifications are delivered, which endpoints are used, nor any access controls or encryption for sensitive categories (financial/legal). That gives the agent broad discretion to send sensitive data over the network with no stated guardrails.
Install Mechanism
This is an instruction-only skill with no install spec and no code files, so it does not write or install third-party artifacts on disk. That minimizes installer-related risk.
Credentials
No environment variables, credentials, or config paths are declared, yet the skill expects cross-agent synchronization and networked notifications. A sync/notification system would typically require service endpoints and credentials (tokens, API keys). The absence of declared credentials is inconsistent and leaves unanswered how the network/API permission will be used safely. The skill also sets a policy of 'audit logs are permanent' which increases data retention risk for sensitive content.
Persistence & Privilege
The skill is not 'always: true' and is user-invocable (defaults), which is normal. It requests file read/write and network API permissions; combined with autonomous invocation, that gives it the capability to read local workspace files and send data over the network. That capability is coherent with a KB but materially increases blast radius if notification endpoints or auth are misconfigured or malicious — the SKILL.md lacks constraints on where/how network calls occur.
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-kb - 安装完成后，直接呼叫该 Skill 的名称或使用
/ai-company-kb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
v1.0.1-en2
Full body English translation
v1.0.1-en
English version
v1.0.1
- 新增支持任务交接（Handoff）协议，可记录和查询标准交接文档。
- 集成 IMA 实时同步中枢，实现共享状态更新自动同步与快照聚合查询。
- 扩展目录结构，增加 handoff/ 任务交接目录，按状态分类存储。
- 优化接口，统一支持写入/读取共享状态、审计日志、交接记录等多类型知识内容。
元数据
常见问题
AI Company KB (EN) 是什么？
AI Company shared knowledge base interface. Unified management of operations records, strategy documents, audit logs. Supports cross-Agent knowledge sharing... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 110 次。
如何安装 AI Company KB (EN)？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install ai-company-kb」即可一键安装，无需额外配置。
AI Company KB (EN) 是免费的吗？
是的，AI Company KB (EN) 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
AI Company KB (EN) 支持哪些平台？
AI Company KB (EN) 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 KB (EN)？
由 JohnSmithfan（@johnsmithfan）开发并维护，当前版本 v1.0.1-en2。
推荐 Skills