Markdown Editor
打开编辑器

AI 编程智能体套件

安全引入编程助手与智能体：智能体规则文件、编码规范、技能与 MCP 服务器、政策与工具权限。

8 个模板，按使用顺序分为 3 步。在编辑器中打开任一模板，填写后可导出为 PDF、DOCX 或 HTML。

第 1 步

问题定义

  1. AGENTS.md Template for AI Coding Agents

    标准

    Repository instructions for AI coding agents: setup, build and test commands, conventions, architecture map, PR and security rules

    AGENTS.md convention (agents.md)

  2. CLAUDE.md Project Memory Template

    标准

    Claude Code project memory: summary, commands, key paths, conventions, testing and workflow rules, gotchas, AGENTS.md import

  3. Cursor Project Rules Template (.mdc)

    精简

    Cursor .cursor/rules/*.mdc files: front matter fields, rule types, a complete example rule and rule ideas

  4. Coding Standards Template

    标准

    Team coding standards: naming, tool-enforced formatting, structure, errors, logging, tests, reviews, language rules

第 2 步

构建

  1. Agent Skill Specification Template (SKILL.md)

    完整

    Agent Skill spec: activation description, SKILL.md frontmatter, layout, scripts, progressive disclosure budget, versioning, security review, evals

    Agent Skills specification (agentskills.io, checked 2026-10-22)

  2. MCP Server Specification Template (Model Context Protocol)

    完整

    Transport, discovery, capabilities, tools, CacheableResult, MRTR confirmation and requestState, OAuth, OWASP 2026 mapping, errors, tests

    Model Context Protocol specification, revision 2026-07-28; OWASP Top 10 for LLM Applications 2026; OWASP Top 10 for Agentic Applications 2026
第 3 步

评审与治理

  1. AI Coding Assistant Policy Template

    完整

    Policy for AI coding assistants and agents: autonomy levels, tool register, secrets, sandboxing, review of generated code, provenance, OWASP ASI and SSDF mapping

    OWASP Top 10 for Agentic Applications 2026; NIST SP 800-218 SSDF 1.1; EU AI Act Art. 4

  2. Agent Tool and Permission Design Template (OWASP ASI02, ASI03)

    完整

    Agent tool permissions: tool inventory, capability classes, least agency, non-human identity, approval tiers, sandboxing, limits, audit, revocation, tests

    OWASP Top 10 for Agentic Applications 2026 (ASI02, ASI03); OWASP Top 10 for LLM Applications 2026 (LLM03 Excessive Agency)