跳到正文
cccskills
免费在 GitHub 公开

omp-roles

Configure tools and skills for sub-agents before dispatching development tasks. Role presets cover implementation, planning, reviews, testing, and verification, with defined boundaries for what each agent should do and whether it may modify files.

中文简介正在准备中。当前显示英语简介。

原文查看原文说明

Use this skill whenever you need the allowed_tools / skills preset for a sub-agent role before calling spawn_subagent. Trigger especially when user mentions: "role", "spawn_subagent", "allowed_tools", or any OMP role name (executor, architect, analyst, critic, security-reviewer, code-reviewer, qa-tester, planner, explore, debugger, verifier). Also use when a /ultrawork, /ralph, /autopilot, /ultraqa, or /team workflow step asks for role tool/skill configuration.

查看安装方法

适用场景

  • Setting tool permissions before spawning a sub-agent
  • Separating implementation, planning, and review roles
  • Configuring agents for review without file changes
  • Assigning role settings within OMP workflows

包含的文件(1)

  • SKILL.md4.1 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

OMP Role Tool Configuration

When using spawn_subagent (single or batch mode) to dispatch sub-agents, set the corresponding allowed_tools and skills parameters based on the sub-agent's role.

Canonical source of truth in code: plugins/bundle/omp_workflows/shared/constants.py (ROLE_ALLOWED_TOOLS, ROLE_SKILLS, TOOL_* names) and shared/role_prompts.py (ROLE_PROMPTS, build_worker_prompt, format_spawn_call). Tool strings must match registered ToolRegistry names — do not invent aliases.

Role Definitions

executor (Code Executor)

  • allowed_tools: null (inherits all tools)
  • skills: null (inherits all skills)
  • Boundary: Implement features, run quality checks. Follow existing patterns.

architect (System Architect)

  • allowed_tools: ["read_file", "grep_search", "glob_search", "write_file", "ast_search", "execute_shell_command"]
  • skills: []
  • Boundary: Design interfaces and module boundaries, diagnose complex issues. No implementation code.

analyst (Requirements Analyst)

  • allowed_tools: ["read_file", "grep_search", "glob_search", "write_file", "execute_shell_command"]
  • skills: []
  • Boundary: Extract requirements, identify constraints, define acceptance criteria. No code.

critic (Plan/Design Critic)

  • allowed_tools: ["read_file", "grep_search", "glob_search", "ast_search"]
  • skills: []
  • Boundary: Review plans and designs, challenge assumptions. No plans, no code.

security-reviewer (Security Reviewer)

  • allowed_tools: ["read_file", "grep_search", "glob_search", "execute_shell_command", "ast_search"]
  • skills: []
  • Boundary: Identify vulnerabilities, injection vectors, auth weaknesses. No file modifications.

code-reviewer (Code Reviewer)

  • allowed_tools: ["read_file", "grep_search", "glob_search", "execute_shell_command", "ast_search"]
  • skills: []
  • Boundary: Review code quality, maintainability, convention adherence. No file modifications.

qa-tester (QA Tester)

  • allowed_tools: ["read_file", "grep_search", "glob_search", "execute_shell_command"]
  • skills: []
  • Boundary: Interactive CLI/service testing — start services and verify behavior.

planner (Strategic Planner)

  • allowed_tools: ["read_file", "grep_search", "glob_search", "write_file", "ast_search"]
  • skills: []
  • Boundary: Create implementation plans, define task breakdown and dependencies. No implementation code.

explore (Code Explorer)

  • allowed_tools: ["read_file", "grep_search", "glob_search", "ast_search"]
  • skills: []
  • Boundary: Read-only exploration. Map codebase structure and dependencies.

debugger (Debugging Expert)

  • allowed_tools: null (inherits all tools)
  • skills: []
  • Boundary: Root cause analysis, fix build/test failures. No new features.

verifier (Adversarial Verifier)

  • allowed_tools: ["read_file", "grep_search", "glob_search", "execute_shell_command", "ast_search"]
  • skills: []
  • Boundary: Verification only — no file modifications. Physical isolation via read-only tool set.

Usage

spawn_subagent(
    task="<role identity from ROLE_PROMPTS>\n\n## Task\n<task description>",
    allowed_tools=<ROLE_ALLOWED_TOOLS[role] or omit if null>,
    skills=<ROLE_SKILLS[role] or omit if null>,
    fork=True,         # if worktree isolation needed
    background=True,   # if background execution needed
)

When allowed_tools or skills is null, omit the parameter (uses default None). For batch mode, pass task="" and put per-item role config in each batch dict.

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Find installation steps, configuration guidance, and command examples for Alibaba Cloud CLI using local Chinese documentation. The skill checks its quick reference or relevant documentation pages before suggesting commands, parameters, and options.

中文简介正在准备中。当前显示英语简介。

  • Look up installation and configuration steps
  • Check commands for querying ECS instances
  • Find output filtering and pagination options
agentscope-ai/QwenPaw3.6万2026年10月22日 更新

browser

免费

Automate browser tasks through QwenPaw's built-in Browser SDK using asynchronous Python. Read page content, fill forms, click controls, and capture screenshots, checking the resulting page after each action to verify what happened.

中文简介正在准备中。当前显示英语简介。

  • Search a website and inspect the results
  • Fill forms and operate page controls
  • Capture page screenshots
agentscope-ai/QwenPaw3.6万2026年10月22日 更新

browser

免费

Use QwenPaw's built-in Browser SDK, a browser control interface, to read pages, search, fill forms, click controls, and capture screenshots through asynchronous Python. The workflow checks the page before and after actions and hands authentication steps to a person.

中文简介正在准备中。当前显示英语简介。

  • Search a website and inspect the resulting page
  • Fill forms and select options in a browser
  • Read page text and capture screenshots
agentscope-ai/QwenPaw3.6万2026年10月22日 更新

Connect to an existing Chrome browser or launch one with an explicit debugging port for sharing across agents and tools. It also covers scanning local CDP ports, the connection endpoints used to control a browser.

中文简介正在准备中。当前显示英语简介。

  • Connect to Chrome with existing logins and open tabs
  • Find available local browser debugging ports
  • Launch a browser with a fixed debugging port
agentscope-ai/QwenPaw3.6万2026年10月22日 更新

Connect an agent to an existing Chrome browser, find local CDP ports used for remote browser control, or start a browser on a specified port for sharing with other agents and tools.

中文简介正在准备中。当前显示英语简介。

  • Connecting to Chrome with existing logins and open tabs
  • Finding available local browser debugging ports
  • Starting a browser on a fixed debugging port
agentscope-ai/QwenPaw3.6万2026年10月22日 更新

Configure how the browser starts, including window visibility, proxy settings, extra launch options, and a custom browser executable. It also explains opening a local debugging port so other local tools can connect.

中文简介正在准备中。当前显示英语简介。

  • Show the browser window during automation
  • Start a browser with a proxy or incognito mode
  • Choose a specific installed browser
agentscope-ai/QwenPaw3.6万2026年10月22日 更新

查看 agentscope-ai 的全部技能

报告此技能的问题