skill-router
Use when the user is unsure which skill to use or where to start. Interviews the user with targeted questions and recommends the best skill(s) from the installed library for their goal.
- License 许可
- MIT
- 最近更新
- 2026-11-10
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- skill-router
- risk
- safe
- source
- self
- description
- Use when the user is unsure which skill to use or where to start. Interviews the user with targeted questions and recommends the best skill(s) from the installed library for their goal.
03 SKILL.md 正文
Skill Router
When to Use
Use this skill when:
- The user says "I don't know where to start" or "which skill should I use"
- The user has a vague goal without a clear method
- The user asks "what should I use for..." or "I'm not sure how to approach this"
- The user is new to the skill library and needs guidance
Goal
Help users who are unsure of what they want to do or which skill to use. Interview them with a short structured conversation, then recommend the most relevant skill(s) from the installed library — with a clear explanation of why each skill fits and exactly how to invoke it.
Instructions
Step 1 — Acknowledge and open the interview
Respond warmly and tell the user you'll ask a few quick questions to find the right skill for them. Do NOT suggest any skills yet.
Example opener:
"No problem — let me ask you a few quick questions so I can point you to exactly the right skill."
Step 2 — Ask the Funnel Questions (one at a time, in order)
Ask only what you need. If an earlier answer makes a later question irrelevant, skip it.
Q1 — What is the broad area of the task? Present these as numbered options:
- Building / coding something (app, feature, component, script)
- Fixing or debugging something that's broken
- Security, pentesting, or vulnerability assessment
- AI agents, LLMs, or automation pipelines
- Marketing, SEO, content, or growth
- DevOps, infrastructure, deployment, or git
- Design, UI/UX, or creative output
- Planning, strategy, or documentation
- Something else (ask them to describe it)
Q2 — How specific is the task?
- I have a clear spec / I know exactly what I want built
- I have a rough idea but need help shaping it
- I'm totally starting from scratch with no clear direction
Q3 — What tech stack or domain is involved? (only ask if relevant) Examples: React / Next.js, Node.js, Python, AWS, Stripe, AI/LLM, no-code, etc. If they say "not sure" or "any", that's fine — move on.
Q4 — Do you want to work autonomously (agent does everything) or collaboratively (you stay in the loop)?
- Fully autonomous — just go
- Collaborative — I want to review/approve steps
- Not sure yet
Step 3 — Recommend skills
Based on their answers, recommend 1 primary skill and up to 2 secondary skills. Structure your recommendation exactly like this:
✅ Primary Skill: @skill-name
Why: [1–2 sentences explaining why this is the best fit for what they described]
Invoke it like this:
@skill-name [paste their goal here]
🔁 Also consider:
@skill-name-2— [one sentence on when to layer this in]@skill-name-3— [one sentence on when to layer this in]
Step 4 — Offer a ready-made prompt
After the recommendation, offer to write them a ready-to-use prompt they can paste directly into Antigravity. Ask:
"Want me to write the full prompt for you so you can just paste it in?"
If yes, compose a complete, specific prompt using the recommended skill and everything you learned about their goal from the interview.
Example Interaction
User: @skill-router I want to build something but I'm not sure where to start
Agent: No problem — let me ask a few quick questions.
What is the broad area?
(1) Building (2) Debugging (3) Security (4) AI/LLM
(5) Marketing (6) DevOps (7) Design (8) Planning
User: 1 - building something
Agent: How specific is the task?
(1) Clear spec (2) Rough idea (3) Starting from scratch
User: 2 - rough idea
Agent: What tech stack? (React, Node, Python, etc. — or "not sure")
User: Next.js
Agent: ✅ Primary Skill: @brainstorming
Why: You have a rough idea that needs shaping before building.
Brainstorming asks structured questions and produces a clear spec.
@brainstorming help me design a [your app idea] using Next.js
🔁 Also consider:
- @plan-writing — once brainstorming produces a spec, break it into tasks
- @senior-fullstack — when you are ready to start building
Want me to write the full prompt for you?
Skill Routing Reference
Building a full product or app from scratch
- Primary:
@app-builder - If they want to plan first:
@brainstorming→@plan-writing→@app-builder - If they want it fully autonomous:
@loki-mode
Building a specific frontend feature / UI
- Primary:
@senior-fullstackor@frontend-design - Stack-specific:
@react-patterns,@nextjs-best-practices,@tailwind-patterns - If they want a full design system:
@ui-ux-pro-max+@core-components
Building a backend API or service
- Primary:
@backend-dev-guidelines - Stack-specific:
@nodejs-best-practices,@python-patterns,@nestjs-expert - API design:
@api-patterns - Database:
@database-design+@prisma-expert
Debugging something broken
- Primary:
@systematic-debugging - If tests are failing:
@test-fixing - If it's a code quality issue:
@clean-code
Writing tests / TDD
- Primary:
@tdd - For Playwright/browser tests:
@playwright-skill - For Jest patterns:
@testing-patterns
Integrating a third-party service
- Payments:
@stripe-integration - Auth:
@clerk-author@nextjs-supabase-auth - Database:
@neon-postgresor@firebase - Messaging:
@twilio-communications - Bots:
@slack-bot-builder,@discord-bot-architect,@telegram-bot-builder - File storage:
@file-uploads - Analytics:
@analytics-tracking
AI / LLM / agents
- Architecture:
@ai-agents-architect - RAG pipelines:
@rag-engineer - Prompts:
@prompt-engineer - Multi-agent:
@langgraphor@crewai - Observability:
@langfuse - Voice:
@voice-agents
Security / pentesting
- Start here:
@ethical-hacking-methodology+@pentest-checklist - Web app testing:
@burp-suite-testing,@sql-injection-testing,@xss-html-injection - Network/infra:
@aws-penetration-testing,@linux-privilege-escalation - Reference:
@top-web-vulnerabilities
DevOps / infrastructure / deployment
- Docker:
@docker-expert - Cloud:
@aws-serverless,@gcp-cloud-run,@vercel-deployment - Git workflow:
@git-pushing,@using-git-worktrees,@github-workflow-automation - Scripting:
@linux-shell-scripting
Marketing / growth / SEO
- Copy:
@copywriting - Landing pages:
@page-cro - SEO:
@seo-fundamentals+@seo-audit - Email:
@email-sequence - Ads:
@paid-ads - Launch:
@launch-strategy
Planning / architecture / strategy
- Quick plan:
@concise-planning - Full plan:
@plan-writing→@executing-plans - Architecture:
@software-architectureor@senior-architect - Product strategy:
@product-manager-toolkit
Creative / design / visuals
- UI:
@frontend-design - Data viz:
@claude-d3js-skill - Generative art:
@algorithmic-art - Presentations:
@pptx-official
Fully autonomous / parallel execution
- Full startup mode:
@loki-mode - Independent parallel tasks:
@dispatching-parallel-agents - Plan then execute:
@subagent-driven-development
Document creation
- Word doc:
@docx-official - PDF:
@pdf-official - Spreadsheet:
@xlsx-official - Presentation:
@pptx-official
Constraints
- Never recommend more than 1 primary skill and 2 secondary skills at a time.
- Always include the exact
@invokesyntax so users can copy-paste it. - If the user's goal spans multiple categories, pick the most upstream skill
(e.g.
@brainstormingbefore@senior-fullstack). - Do not overwhelm the user with the full skill list. Recommend only what is relevant to their specific answers.
- If the user is totally lost, default to
@brainstormingfor open-ended goals, or@app-builderfor anything involving building something. - After recommending, always offer to write a ready-made prompt for them.
Limitations
- Only recommends skills from the installed library. If a skill is not installed, the recommendation may not work.
- Routing is based on natural language matching. Highly ambiguous goals may require follow-up clarification.
- Does not execute the recommended skill — it only recommends it. The user must invoke the skill themselves.
- The routing reference covers the most common skills but does not include every skill in the library.
04 同仓库更多 Skills
vercel-sandbox
Run agent-browser + Chrome inside Vercel Sandbox microVMs for browser automation from any Vercel-deployed app. Use when the user needs browser automation in a Vercel app (Next.js, SvelteKit, Nuxt, Remix, Astro, etc.), wants to run headless Chrome without binary size limits, needs persistent browser sessions across commands, or wants ephemeral isolated browser environments. Triggers include "Vercel Sandbox browser", "microVM Chrome", "agent-browser in sandbox", "browser automation on Vercel", or any task requiring Chrome in a Vercel Sandbox.
slack
Interact with Slack workspaces using browser automation. Use when the user needs to check unread channels, navigate Slack, send messages, extract data, find information, search conversations, or automate any Slack task. Triggers include "check my Slack", "what channels have unreads", "send a message to", "search Slack for", "extract from Slack", "find who said", or any task requiring programmatic Slack interaction.
electron
Automate Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify, etc.) using agent-browser via Chrome DevTools Protocol. Use when the user needs to interact with an Electron app, automate a desktop app, connect to a running app, control a native app, or test an Electron application. Triggers include "automate Slack app", "control VS Code", "interact with Discord app", "test this Electron app", "connect to desktop app", or any task requiring automation of a native Electron application.
dogfood
Systematically explore and test a web application to find bugs, UX issues, and other problems. Use when asked to "dogfood", "QA", "exploratory test", "find issues", "bug hunt", "test this app/site/platform", or review the quality of a web application. Produces a structured report with full reproduction evidence -- step-by-step screenshots, repro videos, and detailed repro steps for every issue -- so findings can be handed directly to the responsible teams.
agent-browser
Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.
shadcn
Manages shadcn components and projects — adding, searching, fixing, debugging, styling, and composing UI, including chat interfaces. Provides project context, component docs, and usage examples. Applies when working with shadcn/ui, component registries, presets, --preset codes, or any project with a components.json file. Also triggers for "shadcn init", "create an app with --preset", "switch to --preset", and for building a design system or applying a DESIGN.md or brand spec to shadcn/ui.