ai-company

微档（XS）AI Agent 组织治理技能包：把 18 个企业职能块归入治理与交付、工程与安全两个部门，随包提供共享方法模板、错误码目录、缩放规范与两套人工作业提示词，并可按阈值评估层级升级、仅生成提案。运行时输出跟随用户语言。

johnsmithfan

热度 38.69

最近更新 2026-09-22 07:45:22

分类：安全依赖：读写文件 / 代码执行 / 内部调用大模型 / MCP 服务应用场景：模板 / 工作流 / 代码生成 / 代码审查 ...

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add johnsmithfan/ai-company

Metadata

name: ai-company
slug: ai-company
version: 1.0.0
description: Micro-tier governance skill for small AI agent organizations (1-3 agents). Consolidates 18 enterprise function blocks into two departments, governance-and-delivery and engineering-and-safety, covering strategy, risk, compliance, architecture, safety, and delivery. Ships ten harness code templates, three prompt frameworks, a full error-code catalog, and a scaling specification. Provides two self-contained human-paste prompts (method implementation, robustness checks) usable in any AI chat window. Source files are authored in English; runtime output follows the user's language per the language policy, with English fallback. Includes propose-only self-scaling that evaluates scaling thresholds and proposes tier upgrades up to group. Use when a small team needs enterprise-grade governance, engineering harnesses, PII masking, AIGC disclosure, or tier-upgrade assessment for its agents.
license: GPL-3.0
author: AI Company Team
tags: ["governance","llm-agents","agent-organization","compliance","error-codes","harness-engineering","pii-masking","aigc-disclosure","self-scaling","micro-tier","method-templates","robustness-checks"]
dependencies: []
triggers: ["evaluate scaling threshold","propose tier upgrade","review governance policy","approve quarterly objectives","audit compliance evidence","assess organizational risk","implement method from specification","apply robustness checks to code","mask sensitive data in outputs","generate trace id for audit trail","review error handling design","respond in user language","translate deliverable content","add aigc disclosure to output","package deliverable for release"]
interface: {"inputs":{"type":"object","schema":{"type":"object","properties":{"task":{"type":"string","description":"Task description"},"department":{"type":"string","enum":["auto","governance-and-delivery","engineering-and-safety"],"description":"Which department to invoke; omit or \"auto\" to auto-route"},"output_language":{"type":"string","description":"Override runtime output language; omit to follow the language decision chain"},"context":{"type":"object","description":"Optional context information"}},"required":["task"]}},"outputs":{"type":"object","schema":{"type":"object","properties":{"result":{"type":"string","description":"Operation result"},"report":{"type":"object","description":"Detailed report data"}},"required":["result"]}}}
errors: []
error_code_prefixes: ["CEO_","CTO_","SCL_"]
language_policy: {"authoring":"en","encoding":"utf-8","output":"user-specified","fallback":"en","immutable_tokens":["slug","error_code","field_name","enum_value","file_path","function_name"]}
self_scaling: {"enabled":true,"state_file":".scaling-state.json","spec":"references/scaling.md","upgrade_policy":"propose-only","max_tier":"group"}
permissions: {"files":{"read":["{WORKSPACE_ROOT}/**","{SKILL_DIR}/**"],"write":["{WORKSPACE_ROOT}/**"],"deny":["~/.ssh/**","~/.aws/**","~/.config/**","/etc/**","{WINDOWS_DIR}/**"]},"network":[],"commands":[],"mcp":["sessions_send","subagents"]}
quality: {"idempotent":true}
metadata: {"category":"enterprise","layer":"AGENT","cluster":"governance","maturity":"STABLE","license":"GPL-3.0","standardized":true,"scale_tier":"micro","department_count":2,"function_block_count":18,"harness_baseline":"L3"}

AI Company Governance

Micro-tier (XS) LLM agent governance skill for organizations of 1-3 agents: 18 enterprise function blocks consolidated into 2 departments, backed by shared harness templates, a full error-code catalog, and propose-only self-scaling.

Source files are authored in English; runtime output follows the user's language (see language_policy in the frontmatter). Slugs, error codes, field names, enum values, file paths, and function names are never translated.

Departments

Slug	Prefix	Scope	Reference
`governance-and-delivery`	`CEO_`	Strategy, governance, delivery	governance-and-delivery.md
`engineering-and-safety`	`CTO_`	Architecture, engineering, safety	engineering-and-safety.md

Shared Resources

Resource	Contents
method-patterns.md	Single source of truth for the ten shared code templates and the CRISPE / 3WEH / Five-Element prompt frameworks
error-codes.md	Full four-element error-code catalog (code, message, trigger, resolution) for all prefixes
scaling.md	Tier thresholds, upgrade gating, and alias mapping for self-scaling

Error Code Prefixes

Prefix	Meaning
`CEO_`	Governance-and-delivery department codes
`CTO_`	Engineering-and-safety department codes
`SCL_`	Self-scaling infrastructure codes (fixed across tiers)

Prompts

File	Mode	Purpose
01-implement-method.md	`human-paste`	Paste into any AI chat window to have that AI implement one method to this standard
02-robustness-checks.md	`human-paste`	Paste into any AI chat window to have that AI run robustness checks on a method

Quick Start

Invoke the skill with task (required) and optionally department (auto routes automatically); set output_language to override the runtime output language.
For implementation standards, use the shared templates in method-patterns.md; for failures, look up codes by prefix in error-codes.md.
To work outside this skill, open 01-implement-method.md or 02-robustness-checks.md, fill in the <placeholders>, and paste the prompt into any AI chat window.
When the organization grows, ask the skill to evaluate scaling thresholds or propose a tier upgrade; the rules live in scaling.md.