功能描述

Unified AI Company skill consolidating 16 department skills into one. Provides complete governance, finance, technology, security, legal, people, marketing,...

使用说明 (SKILL.md)

\r \r

AI Company v1.1.0\r

Name: ai-company-unified
Author: johnsmithfan

\r

Unified AI Company Skill — 16 departments consolidated into one.\r Full specifications in references/method-patterns.md and references/departments/.\r \r

Quick Reference\r

\r

What This Skill Does\r

Complete AI company operations: governance, finance, technology, security, legal, people, marketing, quality, intelligence, information, translation, and platform infrastructure. Use for any AI-Company function.\r \r

Department Index\r

\r | Department | Roles | Details |\r |-----------|--------|---------|\r | Governance & Strategy | CEO, COO, HQ | governance-and-strategy.md |\r | Finance & Risk | CFO, CRO | finance-and-risk.md |\r | Technology & Engineering | CTO | technology-and-engineering.md |\r | Platform & Infrastructure | Framework | platform-and-infrastructure.md |\r | Security & Compliance | CISO, CLO | security-and-compliance.md |\r | People & Culture | CHO | people-and-culture.md |\r | Marketing & Partnerships | CMO | marketing-and-partnerships.md |\r | Quality & Operations | CQO, PMGR | quality-and-operations.md |\r | Intelligence | Intel | intelligence.md |\r | Information Services | Information | information.md |\r | Translation & Localization | Translator | translation-and-localization.md |\r \r

Shared Resources\r

\r

Error Codes\r

\r All error codes use department prefix (e.g., CEO_001, CFO_001, CISO_001). See individual department files for complete error code reference and resolution steps.\r \r

Prompts\r

\r Copy-paste ready prompts in prompts/:\r

Auto-Update\r

\r This skill supports automatic updates from ClawHub with 5-layer security gates.\r \r | Setting | Value |\r |---------|-------|\r | Schedule | Weekly Sunday 02:00 UTC |\r | RRule | FREQ=WEEKLY;BYDAY=SU;BYHOUR=2;BYMINUTE=0 |\r | Backup Retention | 10 versions / 30 days |\r \r Security Gates: Version Check | Backup Gate | Download Gate | Frontmatter Gate | Danger Pattern Gate\r \r Manual Update:\r

pwsh -File "C:\Users\Admin\WorkBuddy\Claw\.workbuddy\scripts\ai-company-auto-update.ps1" -Force\r
```\r
\r
**Logs**: `C:\Users\Admin\WorkBuddy\Claw\.workbuddy\logs\ai-company-update-log.md`\r
\r
## Changelog\r
\r
| Version | Date | Changes |\r
|---------|------|---------|\r
| 1.0.4 | 2026-04-29 | Intel: Added Intelligence Library (SOP-L01~L06) with auto-triggered library setup on first collection request; SOP-L01 auto-creates library silently without user prompt; SOP-L06 triggers on any intelligence collection request (收集情报/intelligence collection); Added INTEL_006~INTEL_010 error codes; Updated triggers with 情报收集/intelligence collection/intelligence library |\r
| 1.0.3 | 2026-04-28 | Security: Scoped file permissions to WORKSPACE_ROOT (P0 CISO fix); Finance: Added capex policy, working capital DSO/DPO targets, CRO-CFO escalation SLA (P1 CFO/CRO); Risk: Added numeric FAIR thresholds and LEA calculation (P1 CRO); CTO: Added 3-stage deployment gate with rollback triggers (P1); CQO: Added test coverage acceptance threshold 85% (P1); CEO: Added board escalation ladder (P2); COO: Added automated OHS alerting + OKR integration in MEASURE phase (P2); CLO: Added DMCA takedown workflow (P2); Intel: Added 6-phase intelligence cycle (P2); CPO: Added semver enforcement policy (P2) |\r
| 1.0.2 | 2026-04-27 | Added auto-update: weekly automation (Sunday 02:00 UTC), PowerShell script with 5-layer security gates, backup/rollback, update log, publisher allowlist |\r
| 1.0.1 | 2026-04-27 | CEO review complete: all 7 reference modules verified and rebuilt; added visualization.md, integrations.md, memory.md, data-integration.md, execution.md |\r
| 1.0.0 | 2026-04-27 | Initial release to ClawHub as unified AI Company skill; 16 departments consolidated |\r
\r
---\r
\r
*This skill follows AI Company Governance Framework. See [references/](references/) for complete specifications.*

安全使用建议

This is a very large, all-in-one policy and workflow pack that asks an agent to design and execute cross-department actions (including agent creation, CI/CD, and crisis responses). Before installing: 1) Recognize scope: it gives the agent wide operational guidance — avoid enabling autonomous runs on production systems. 2) Clarify external services: ask the author how VirusTotal/ClawHub scanning is performed and where API keys should be stored; prefer explicitly declared env vars/scopes rather than ad-hoc keys. 3) Least privilege: if you test it, run in a sandboxed agent with no access to cloud credentials, SSH keys, or production deploy tooling. 4) Audit & logging: require that any real-world actions (deploys, permission changes, shutdowns) have human approval steps and are logged/audited. 5) Review docs: verify the L1-L6 harness and the 'no eval/exec' guarantees are enforced by policy rather than just in prose. If you need this functionality, request the publisher to: (a) declare required credentials/env vars explicitly, (b) remove/limit examples that imply issuing system-wide commands, and (c) add explicit gating (human-in-loop) for any destructive or privileged actions.

功能分析

Type: OpenClaw Skill Name: ai-company-unified Version: 1.0.1 The skill bundle is a highly comprehensive and structured framework designed to simulate an entire AI-driven corporation across 16 departments. It includes extensive documentation on security protocols (STRIDE, CVSS), quality gates (G0-G7), and defensive engineering standards (L1-L6). The provided code templates in 'references/departments/platform-and-infrastructure.md' are explicitly designed with security in mind, featuring input sanitization, PII masking, and sandboxed command execution with timeout enforcement. The instructions for the AI agent consistently reinforce safety boundaries, prohibiting the use of eval/exec and unauthorized access to sensitive system paths like ~/.ssh. No evidence of data exfiltration, malicious execution, or obfuscation was detected.

能力标签

cryptocan-make-purchasesrequires-oauth-tokenrequires-sensitive-credentials

能力评估

ℹ Purpose & Capability

The name/description claim to consolidate many departments — governance, finance, security, legal, platform, etc. — which matches the included department documents and prompts. However the SKILL.md and prompts reference external verification services (VirusTotal/ClawHub) and multi-source integrations that usually require credentials or installers; yet the skill declares no required env vars or install steps. That mismatch (mentioning third-party verification but not declaring API keys or how to access those services) is unexpected and worth questioning.

⚠ Instruction Scope

The runtime instructions ask agents to "implement" department functions, run pre/post compliance checks, and perform workflows including agent creation, CI/CD, production deployment, and crisis actions (examples include system-wide shutdown/restart commands and temporary permission elevation in a crisis whitelist). Although the prompts also state 'No eval/exec/remote loading' and include security checks, the instruction language gives broad operational authority and discretionary power to an agent. The skill bundle also contains explicit test cases that reference reading sensitive paths (e.g., /etc/passwd) and checks that such accesses be blocked — implying the agent may otherwise attempt system file accesses. This broad, action-oriented scope with few concrete limits is a red flag.

✓ Install Mechanism

This is instruction-only with no install spec or code files to execute. That minimizes direct installation risk (no archives to download, no post-install scripts).

ℹ Credentials

The skill declares no required environment variables or credentials, but multiple reference documents mention optional API keys (e.g., GOOGLE_GEOLOCATION_API_KEY, MLS_API_KEY) and external services (VirusTotal) that typically require credentials. The absence of declared required env vars is inconsistent with the functionality described and may lead implementers to supply sensitive keys ad-hoc. The skill also references integration with internal systems (HQ, Harness, agent factory) which in practice would require credentials or privileged access — none are declared.

✓ Persistence & Privilege

always: false and no OS restrictions; the skill does not request forced or persistent inclusion and does not declare modifications to other skills or system-wide config. Autonomous invocation is enabled by default (platform normal), but there is no evidence the skill demands elevated installation privileges.

版本历史

v1.0.3

Security: Scoped file permissions to WORKSPACE_ROOT (P0 CISO fix); Finance: Added capex policy, working capital DSO/DPO targets, CRO-CFO escalation SLA; Risk: Added FAIR LEA thresholds; CTO: Added 3-stage deployment gate; CQO: Test coverage threshold 85%; CEO: Board escalation ladder; COO: OHS alerting + OKR integration; CLO: DMCA takedown workflow; Intel: 6-phase intelligence cycle; CPO: Semver enforcement policy

v1.0.1

CEO review: 5 missing modules rebuilt; v8.0.0 full reference suite; 1.0.1 patch release

v1.0.0

Initial release: Unified AI Company skill with 16 departments consolidated into one; 11 department reference files, shared code templates, prompt frameworks (CRISPE/3WEH/Five-Element), and compliance verification.

v5.0.0

Consolidated 16 AI-Company department skills into single unified skill. 11 department files, 10 code templates, 3 prompt frameworks (CRISPE/3WEH/Five-Element), L1-L6 harness engineering, AIGC compliance, VirusTotal/ClawHub safety verified.

元数据

Slug ai-company-unified

版本 1.0.1

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 3

常见问题

ai-company-unified 是什么？

Unified AI Company skill consolidating 16 department skills into one. Provides complete governance, finance, technology, security, legal, people, marketing,... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 30 次。

如何安装 ai-company-unified？

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

ai-company-unified 是免费的吗？

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

ai-company-unified 支持哪些平台？

ai-company-unified 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-unified？

由 JohnSmithfan（@johnsmithfan）开发并维护，当前版本 v1.0.1。

ai-company-unified