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.
- 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
- 参考文档references/architecture_reviewer.md1.1 KB
- 参考文档references/code_reviewer.md998 B
- 参考文档references/compatibility_reviewer.md1.5 KB
- 参考文档references/design_reviewer.md1.6 KB
- 参考文档references/devops_reviewer.md993 B
- 参考文档references/doc_reviewer.md904 B
- 参考文档references/security_reviewer.md973 B
- 参考文档references/test_reviewer.md937 B
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
-
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.
-
Load Personas: For each relevant domain, read the corresponding reference file in
references/:references/code_reviewer.mdreferences/architecture_reviewer.mdreferences/security_reviewer.mdreferences/design_reviewer.mdreferences/doc_reviewer.mdreferences/test_reviewer.mdreferences/devops_reviewer.mdreferences/compatibility_reviewer.md
-
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). -
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_investigatororgeneralistsub-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
good-skill
A well-formed fixture skill.
some-other-name
Name does not match its directory.
no-desc
暂无简介
summon-profile
Synchronizes the local Gemini profile (~/.gemini) with a remote machine via scp.
start-worktree
Creates a new isolated git worktree for a JIRA ticket, generates a draft PR checkpoint doc, and sets up the branch following user conventions. Use when starting a new task or feature.
release
Automates the version bump, changelog, build, and git release process for the repository. Use this when the user is ready to cut a new release of the toolkit.