菜单
Design GitHub Copilot-based agentic AI systems. Use when turning a user brief into a complete system with agents, prompts, skills, instructions, hooks, MCP integration, manifests, runbooks, and rollout guidance.
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。
基于 SOC 职业分类
Parse AUTOSAR, MISRA, and similar C++ standards documents, extract rule records, map rule interactions, and synthesize a core guideline with canonical coding patterns and fix-order guidance.
Use when identifying and modeling task dependencies in a project plan. Covers building a Directed Acyclic Graph (DAG) of task relationships using Mermaid flowchart syntax, detecting dependency cycles, and computing the critical path. Applies to any task list that needs a structured execution order.
Use when identifying, classifying, and mitigating project risks. Covers the full risk management workflow: brainstorming risks across five categories, scoring them on a probability × impact matrix, writing proactive mitigation strategies, reactive contingency plans, and maintaining a Markdown risk register.
Use when breaking a complex goal or project requirement into atomic, measurable, assignable tasks. Covers the full SMART decomposition workflow: understanding the goal, identifying deliverables, creating work packages, assigning ownership and duration, and validating completeness. Applies to software projects, planning agents, and any structured task list creation.
Use when asked to create a new local MCP (Model Context Protocol) tool/server for GitHub Copilot — file readers, API wrappers, CLI integrations, or any capability that Copilot cannot do natively. Covers the full workflow: Python MCP server implementation, dependency setup, VS Code mcp.json registration, Windows installation script, and end-to-end verification. Applies to any OS-local tool that must run without internet access (corporate environments).
Use when asked to generate a git commit message for a code change. Reads the diff (unstaged or staged), infers the commit type and scope from the changed files, then produces a commit message that matches the project's convention — either from a templates/commit-template.md file in this skill folder, or derived from the last 5–10 commits in the repo.
| name | agentic-system-factory |
| description | Design GitHub Copilot-based agentic AI systems. Use when turning a user brief into a complete system with agents, prompts, skills, instructions, hooks, MCP integration, manifests, runbooks, and rollout guidance. |
| argument-hint | Describe the target system, domain, infrastructure, integrations, constraints, and success criteria |
| user-invocable | true |
Use this skill to design complete agentic AI systems on top of GitHub Copilot.
.github/ customization bundle planGather or confirm the following whenever possible:
references/copilot-bundle-template.md.references/system-blueprint-template.mdreferences/copilot-bundle-template.mdreferences/implementation-roadmap-template.md