swe-re-expert

Usage Guide: Compose and Use an AAS Skill Stack

Recommended path: let Codex or Claude inspect the project, search and read the complete AAS catalog, and choose the exact skills. AAS Core records and validates that agent-owned selection.

Primary workflow: agent-first composition

After configuring the local AAS MCP, ask your agent to inspect the repository and choose a stack for the outcome you want:

Inspect this repository and enumerate its primary capability areas. For each
capability, search the complete AAS catalog, paginate or refine the query, and
compare multiple plausible candidates with get_skill when available. Select at
least one non-redundant valid skill per capability, explicitly report catalog
gaps, and evaluate architecture/runtime, languages/frameworks, domain behavior,
data/storage, integrations, testing/quality, security/privacy, UX/accessibility,
deployment/operations, and maintenance workflow. Mark dimensions not applicable
instead of silently omitting them. Do not stop at the first few matches or
optimize for the smallest stack. Core has no semantic policy favoring a small
stack; each manifest has a technical maximum of 128 selected skills. Only then
use compose_stack with a project profile, inspect the schema 2 manifest returned
in memory, and do not apply it.

The agent must use search_skills and get_skill across the complete catalog, build a capability-to-skill coverage map, continue searching while a primary capability remains uncovered, choose the exact IDs itself, call compose_stack, then check the in-memory proposal with inspect_stack before presenting it. Every current catalog skill remains individually searchable, readable, and selectable. A client or the CLI can persist the reviewed aas-stack.json; an audit-enabled flow can then call export_selection_evidence, validate it with inspect_selection_evidence, and atomically publish the manifest and separate aas-selection-evidence.json sidecar in an artifact-dir. Use aas stack plan to preview the exact operations without materializing skills or managed state in the target.

Selection evidence contains the raw search_skills queries observed during the MCP session. Do not place secrets, credentials, private source text, or personal data in catalog queries.

Selection belongs to the coding agent. AAS MCP does not inspect the repository itself, rank or exclude skills, install skills, update catalogs, or change configuration. See AAS Core for setup, the exact tool boundary, CLI commands, and preview limitations.

stack apply and stack recover are experimental, disabled by default, and are not supported or certified preview safety claims.

Alternative workflow: direct skill installation

If you came in through a Claude Code or Codex plugin instead of AAS Core or a full library install, invoke individual skills in prompts. Plugins ship a fixed plugin-safe subset; AAS Core instead validates and records the exact stack the agent selected from the verified catalog in aas-stack.json. See plugins.md for the distribution model.

What direct distribution provides

When you ran npx agentic-awesome-skills or cloned the repository, you:

✅ Downloaded 2,715+ skill files to your computer (default: ~/.agents/skills/; or a custom path like ~/.agent/skills/ if you used --path) ✅ Made them available to your AI assistant ❌ Did NOT enable them all automatically (they’re just sitting there, waiting)

Direct distribution makes skill files available to the host. It does not select a project-specific stack, record desired state, or produce a preview plan; those are AAS Core responsibilities.

Direct-install Step 1: Understanding Bundles

Common confusion: “Do I need to download each skill separately?”

Answer: NO! You do not need to download each skill separately. Here’s what bundles actually are:

What Bundles Are

Bundles are curated groups of skills organized by role. They help you decide which skills to start using, and they can also be exposed as focused marketplace plugins for Claude Code and Codex.

Bundles provide editorial shortlists. You can either select individual skills from a bundle or install its focused marketplace plugin where supported.

What Bundles Are NOT

❌ Separate skill downloads ❌ Invokable mega-skills like @essentials or /web-wizard ❌ Something most users need to activate during normal install ❌ A replacement for invoking the individual skills inside the bundle

Example: The “AAS Web App Builder” Plugin

When you see the AAS Web App Builder pack, it lists:

These are recommendations for which skills a web developer should try first. If you have the full library installed, you just need to use them in your prompts. If you prefer a narrower install surface, you can install the matching bundle plugin in Claude Code or Codex where plugin marketplaces are available.

The key distinction is:

See plugins.md for the canonical explanation.

If you want only one bundle active at a time in Antigravity, use the activation scripts instead of trying to invoke the bundle name directly:

./scripts/activate-skills.sh --clear Essentials
./scripts/activate-skills.sh --clear "Web Wizard"

Direct-install Step 2: Invoke a Skill

The Simple Answer

Just mention the skill name in your conversation with your AI assistant.

Different Tools, Different Syntax

The exact syntax varies by tool, but it’s always simple:

Claude Code (CLI)

# In your terminal/chat with Claude Code:
>> Use @brainstorming to help me design a todo app

Cursor (IDE)

# In the Cursor chat panel:
@brainstorming help me design a todo app

Gemini CLI

# In your conversation with Gemini:
Use the brainstorming skill to help me plan my app

If Gemini CLI starts hanging after a few turns, try a fresh conversation and temporarily reduce the active set to just 2-5 skills to rule out context growth.

Codex CLI

# In your conversation with Codex:
Apply @brainstorming to design a new feature

Antigravity IDE

# In agent mode:
Use @brainstorming to plan this feature

Pro Tip: Most modern tools use the @skill-name syntax. When in doubt, try that first!

Direct-install Step 3: Write a Focused Prompt

Here are real-world examples of good prompts:

Example 1: Starting a New Project

Bad Prompt:

“Help me build a todo app”

Good Prompt:

“Use @brainstorming to help me design a todo app with user authentication and cloud sync”

Why it’s better: You’re explicitly invoking the skill and providing context.

Example 2: Reviewing Code

Bad Prompt:

“Check my code”

Good Prompt:

“Use @lint-and-validate to check src/components/Button.tsx for issues”

Why it’s better: Specific skill + specific file = precise results.

Example 3: Security Audit

Bad Prompt:

“Make my API secure”

Good Prompt:

“Use @api-security-best-practices to review my REST endpoints in routes/api/users.js”

Why it’s better: The AI knows exactly which skill’s standards to apply.

Example 4: Combining Multiple Skills

Good Prompt:

“Use @brainstorming to design a payment flow, then apply @stripe-integration to implement it”

Why it’s good: You can chain skills together in a single prompt!

Direct-install Step 4: Your First Skill

Let’s actually use a skill right now. Follow these steps:

Scenario: You want to plan a new feature

  1. Pick a skill: Let’s use brainstorming (from the “Essentials” bundle)

  2. Open your AI assistant (Claude Code, Cursor, etc.)

  3. Type this exact prompt:

    Use @brainstorming to help me design a user profile page for my app

  4. Press Enter

  5. What happens next:
    • The AI loads the brainstorming skill
    • It will start asking you structured questions (one at a time)
    • It will guide you through understanding, requirements, and design
    • You answer each question, and it builds a complete spec
  6. Result: You’ll end up with a detailed design document—without writing a single line of code yet!

Direct-install Step 5: Pick Skills Manually

Don’t try to use all 2,715+ skills at once. Here’s a sensible approach:

If you want a tool-specific starting point before choosing skills, use:

Start with “The Essentials” (5 skills, everyone needs these)

  1. @brainstorming - Plan before you build
  2. @lint-and-validate - Keep code clean
  3. @git-pushing - Save work safely
  4. @systematic-debugging - Fix bugs faster
  5. @concise-planning - Organize tasks

How to use them:

Then Add Role-Specific Skills (5-10 more)

Find your role in bundles.md and pick 5-10 skills from that bundle.

Example for Web Developer:

Example for Security Engineer:

Finally, Add On-Demand Skills (as needed)

Keep the CATALOG.md open as reference. When you need something specific:

“I need to integrate Stripe payments” → Search catalog → Find @stripe-integration → Use it!

Complete Example: Building a Feature End-to-End

Let’s walk through a realistic scenario:

Task: “Add a blog to my Next.js website”

Step 1: Plan (use @brainstorming)

You: Use @brainstorming to design a blog system for my Next.js site

AI: [Asks structured questions about requirements]
You: [Answer questions]
AI: [Produces detailed design spec]

Step 2: Implement (use @nextjs-best-practices)

You: Use @nextjs-best-practices to scaffold the blog with App Router

AI: [Creates file structure, sets up routes, adds components]

Step 3: Style (use @tailwind-patterns)

You: Use @tailwind-patterns to make the blog posts look modern

AI: [Applies Tailwind styling with responsive design]

Step 4: SEO (use @seo-audit)

You: Use @seo-audit to optimize the blog for search engines

AI: [Adds meta tags, sitemaps, structured data]

Step 5: Test & Deploy

You: Use @test-driven-development to add tests, then @vercel-deployment to deploy

AI: [Creates tests, sets up CI/CD, deploys to Vercel]

Result: Professional blog built with best practices, without manually researching each step!

Common Questions

“Which tool should I use? Claude Code, Cursor, Gemini?”

Any of them! Skills work universally. Pick the tool you already use or prefer:

“Can I see all available skills?”

Yes. With AAS Core, ask the agent to call search_skills and inspect candidates with get_skill. On a direct install, you can also:

  1. Browse CATALOG.md (searchable list)
  2. Run ls ~/.agents/skills/ (or your actual install path)
  3. Ask your AI: “What skills do you have for [topic]?”

“Do I need to restart my IDE after installing?”

Usually no, but if your AI doesn’t recognize a skill:

  1. Try restarting your IDE/CLI
  2. Check the installation path matches your tool
  3. Try the explicit path: npx agentic-awesome-skills --claude (or --cursor, --gemini, etc.)

“Can I load all skills into the model at once?”

No. Even though you have 2,715+ skills installed locally, you should not concatenate every SKILL.md into a single system prompt or context block.

The intended pattern is:

If you are building your own host/agent (e.g. Jetski/Cortex + Gemini), see:

The v1 manifest shape is defined in:

“Can I create my own skills?”

Yes! Use the @skill-creator skill:

Use @skill-creator to help me build a custom skill for [your task]

“What if a skill doesn’t work as expected?”

  1. Check the skill’s SKILL.md file directly in your installed path, for example: ~/.agents/skills/[skill-name]/SKILL.md
  2. Read the description to ensure you’re using it correctly
  3. Open an issue with details

Quick Reference Card

Save this for quick lookup:

Task Skill to Use Example Prompt
Plan new feature @brainstorming Use @brainstorming to design a login system
Review code @lint-and-validate Use @lint-and-validate on src/app.js
Debug issue @systematic-debugging Use @systematic-debugging to fix login error
Security audit @api-security-best-practices Use @api-security-best-practices on my API routes
SEO check @seo-audit Use @seo-audit on my landing page
React component @react-patterns Use @react-patterns to build a form component
Deploy app @vercel-deployment Use @vercel-deployment to ship this to production

Next Steps

For the Core-first path:

  1. Configure the local MCP with the AAS Core guide.
  2. Ask the agent to search the complete catalog, choose exact IDs, and explain its selection.
  3. Review aas-stack.json, validate it, and preview the plan.

For direct/manual use:

  1. ✅ Try one skill right now - Start with @brainstorming on any idea you have
  2. 📚 Pick 3-5 skills from your role’s bundle in bundles.md
  3. 🔖 Bookmark CATALOG.md for when you need something specific
  4. 🎯 Try a workflow from workflows.md for a complete end-to-end process

Pro Tips for Maximum Effectiveness

Tip 1: Start Every Feature with @brainstorming

Before writing code, use @brainstorming to plan. You’ll save hours of refactoring.

Tip 2: Chain Skills in Order

Don’t try to do everything at once. Use skills sequentially: Plan → Build → Test → Deploy

Tip 3: Be Specific in Prompts

Bad: “Use @react-patterns” Good: “Use @react-patterns to build a modal component with animations”

Tip 4: Reference File Paths

Help the AI focus: “Use @security-auditor on routes/api/auth.js”

Tip 5: Combine Skills for Complex Tasks

“Use @brainstorming to design, then @test-driven-development to implement with tests”

Still Confused?

If something still doesn’t make sense:

  1. Check the FAQ
  2. See Real-World Examples
  3. Open a Discussion
  4. File an Issue to help us improve this guide!

Remember: You’re not alone! The whole point of this project is to make AI assistants easier to use. If this guide didn’t help, let us know so we can fix it. 🙌

This site is open source. Improve this page.