菜单
Agentic AI and intelligent automation knowledge for consulting engagements. Use when working on process automation, multi-agent systems, RPA/IPA implementation, workflow automation, or custom AI solutions. Covers automation assessment, agent architecture, and implementation approaches.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/HossamDaoud83/CPS-Plugins-Official --skill agentic-ai命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
基于 SOC 职业分类
Initialise a new film production from an idea — refresh live pricing, run the intake interview including budget and quality tier, produce a costed plan, and scaffold only after approval. Trigger - /film-setup, once at the start of a new film.
Re-measure every model's live credit cost and rewrite the cost matrix. MANDATORY at the start of every new film project, before quoting or planning. Trigger - /pricing-refresh, or automatically as the first step of film-setup.
Cut, grade, title and master the film locally with ffmpeg — zero credits. Trigger - /assembly, after all clips render and before qc-gate.
| name | agentic-ai |
| description | Agentic AI and intelligent automation knowledge for consulting engagements. Use when working on process automation, multi-agent systems, RPA/IPA implementation, workflow automation, or custom AI solutions. Covers automation assessment, agent architecture, and implementation approaches. |
Provides specialized knowledge for automation and agentic AI engagements including opportunity assessment, agent design patterns, platform selection, and implementation approaches.
| Criterion | Weight | Score (1-5) | Description |
|---|---|---|---|
| Volume | 20% | - | Transaction frequency |
| Standardization | 20% | - | Rule-based vs. judgment |
| Data Structure | 15% | - | Structured vs. unstructured |
| Stability | 15% | - | Process change frequency |
| Error Rate | 15% | - | Current quality issues |
| Strategic Value | 15% | - | Business criticality |
High Volume
│
┌───────────────────┼───────────────────┐
│ AUTOMATE NOW │ OPTIMIZE FIRST │
│ (Quick Wins) │ (Standardize) │
High├───────────────────┼───────────────────┤
Rule│ │ │
Base│ EVALUATE ROI │ LOW PRIORITY │
│ (Case-by-case) │ (Manual OK) │
└───────────────────┼───────────────────┘
Low Volume
Annual Benefit = (Hours Saved × Hourly Cost) + Error Reduction Value + Speed Value
Implementation Cost = License + Development + Training + Change Management
ROI = (Annual Benefit - Annual Cost) / Implementation Cost × 100
Payback Period = Implementation Cost / Annual Benefit
User Request
│
▼
┌─────────────────────────────────────┐
│ AGENT │
│ ┌─────────────────────────────┐ │
│ │ Reasoning Engine │ │
│ │ (LLM + System Prompt) │ │
│ └─────────────────────────────┘ │
│ │ │
│ ┌───────────┴───────────┐ │
│ ▼ ▼ ▼ │
│ Tool 1 Tool 2 Tool 3 │
│ (API) (Database) (Search) │
└─────────────────────────────────────┘
│
▼
Response
Supervisor Pattern:
┌──────────────┐
│ Supervisor │
│ Agent │
└──────┬───────┘
┌───────────┼───────────┐
▼ ▼ ▼
┌────────┐ ┌────────┐ ┌────────┐
│ Worker │ │ Worker │ │ Worker │
│ Agent 1│ │ Agent 2│ │ Agent 3│
└────────┘ └────────┘ └────────┘
Pipeline Pattern:
Input → Agent 1 → Agent 2 → Agent 3 → Output
(Extract) (Transform) (Validate)
Collaborative Pattern:
┌────────────────────────────────────┐
│ Shared Context │
└────────────────────────────────────┘
▲ ▲ ▲
│ │ │
┌────────┐ ┌────────┐ ┌────────┐
│ Agent 1│◄──►│ Agent 2│◄──►│ Agent 3│
└────────┘ └────────┘ └────────┘
| Type | Technologies | Best For |
|---|---|---|
| Task Automation | UiPath, Automation Anywhere, Power Automate | Desktop/web tasks |
| Workflow Automation | Camunda, Temporal, n8n | Process orchestration |
| Document Processing | ABBYY, Kofax, Azure Form Recognizer | Unstructured data |
| Integration | MuleSoft, Workato, Zapier | System connectivity |
| AI/ML | OpenAI, Anthropic, Hugging Face | Intelligent automation |
| Framework | Strengths | Best For |
|---|---|---|
| LangChain | Ecosystem, flexibility | General purpose |
| Claude Agent SDK | Anthropic integration, simplicity | Claude-based agents |
| AutoGen | Multi-agent, Microsoft | Complex workflows |
| CrewAI | Role-based, simple | Team simulations |
| Semantic Kernel | Microsoft stack | Enterprise .NET |
| Category | Examples |
|---|---|
| Finance | Invoice processing, reconciliation, reporting |
| HR | Onboarding, benefits administration, reporting |
| IT | Ticket routing, user provisioning, monitoring |
| Operations | Order processing, inventory updates, shipping |
| Customer Service | Query routing, FAQs, status updates |
| Compliance | Data validation, reporting, audit prep |
| Category | Examples |
|---|---|
| Research | Information gathering, summarization |
| Analysis | Data analysis, insight generation |
| Content | Document creation, communication drafts |
| Support | Customer inquiries, internal helpdesk |
| Coding | Code generation, review, debugging |
| Operations | Monitoring, alerting, remediation |
| Risk | Mitigation |
|---|---|
| Scope creep | Fixed scope per phase, change control |
| Integration failures | Thorough API testing, fallback procedures |
| Data quality | Input validation, exception handling |
| User adoption | Change management, training, champions |
| Maintenance burden | Documentation, COE ownership |
| Control | Implementation |
|---|---|
| Output filtering | Content moderation, format validation |
| Action limits | Scope restrictions, approval gates |
| Rate limiting | Request throttling, cost controls |
| Audit logging | Complete action history |
| Human escalation | Confidence thresholds, sensitive triggers |
| Metric | Target |
|---|---|
| Process time reduction | >70% |
| Error rate reduction | >90% |
| Cost per transaction | >50% reduction |
| Employee satisfaction | Increase |
| Automation coverage | % of eligible processes |
| Metric | Target |
|---|---|
| Task completion rate | >95% |
| Accuracy | >98% |
| Response time | <X seconds |
| Escalation rate | <5% |
| User satisfaction | >4.0/5.0 |
See references/ folder for: