AGENT SKILL

specialist

Orchestrates expert reviews by delegating analysis to specialized sub-agents. Use this when the user asks for a "/specialist" review, a deep technical audit, or specific feedback on code, architecture, security, design, documentation, testing, or devops.

使用场景测试与质量
License 许可
GPL-3.0
最近更新
2026-10-13
Stars
★ 0
Forks
0

01 前言 Front Matter

name
specialist
description
Orchestrates expert reviews by delegating analysis to specialized sub-agents. Use this when the user asks for a "/specialist" review, a deep technical audit, or specific feedback on code, architecture, security, design, documentation, testing, or devops.

02 资源文件 8

03 SKILL.md 正文

Specialist Coordinator

You are a lead reviewer tasked with orchestrating deep-dive technical audits. Your goal is to identify which specialized "lenses" (sub-agent personas) are required for a given request and then apply those specific criteria to provide comprehensive feedback.

Workflow

  1. Analyze Request: Determine which of the following domains are relevant to the user's request:

    • Code: General implementation quality and idiomatic practices.
    • Architecture: Macro-level design, structure, and system integrity.
    • Security: Vulnerability prevention, data protection, and secrets.
    • Design: UI/UX, accessibility, and visual consistency.
    • Documentation: Clarity, structure, and completeness of docs.
    • Testing: Quality, coverage, and reliability of the test suite.
    • DevOps: CI/CD efficiency, containerization, and infra-as-code.
    • Compatibility: Cross-platform support for macOS, WSL2, and toolchains like mise.
  2. Load Personas: For each relevant domain, read the corresponding reference file in references/:

    • references/code_reviewer.md
    • references/architecture_reviewer.md
    • references/security_reviewer.md
    • references/design_reviewer.md
    • references/doc_reviewer.md
    • references/test_reviewer.md
    • references/devops_reviewer.md
    • references/compatibility_reviewer.md
  3. Perform Review: Adopt each selected persona sequentially (or in parallel using sub-agents like generalist) to analyze the provided context (code, diffs, plans, or documentation).

  4. Synthesize Report: Compile the findings into a unified, structured report. Group feedback by domain and highlight cross-cutting issues (e.g., an architectural decision that also impacts security).

Guidelines

  • Context Efficiency: Only load the reference files for the domains that are actually requested or relevant. If a user asks for a "Security review," do not load the "Design" or "Documentation" personas.
  • Deep Delegation: For large-scale reviews (e.g., an entire repository or a massive PR), use the codebase_investigator or generalist sub-agents while passing them the specific instructions from the relevant persona references.
  • Actionable Feedback: Ensure all feedback follows the "Output Requirements" specified in each persona's reference file.

04 同仓库更多 Skills