961882

961882 / ray-skills

Public

一组面向内容选题规划、多角色讨论和角色库扩展的 AI Agent Skills，支持 Codex、Claude Code 和 Cursor 使用。

25
5
89% credibility
Found Jun 02, 2026 at 25 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
AI Summary

Ray Skills is a framework that transforms a single AI assistant into a whole team of virtual experts. Users select from a roster of specialized roles — like a business analyst, system architect, developer, or product manager — and choose how these experts should interact: debating fiercely, collaborating calmly, or converging on a decision. The system orchestrates realistic multi-party conversations where each role contributes their unique perspective, challenges assumptions, and builds toward a richer answer than any single assistant could provide. It supports both product-development scenarios and content-creation workflows.

How It Works

1
💡 You discover a smarter way to brainstorm

You hear about a tool that lets AI assistants take on different expert roles — like having a whole team of specialists thinking together.

2
🎭 You pick your dream team of experts

Choose from a cast of specialists — a business analyst who digs into market research, an architect who weighs trade-offs, a developer who cares about clean code, and many more.

3
You choose how they should work together
⚔️
Debate mode

Experts challenge each other to stress-test every idea

🤝
Roundtable mode

Everyone collaborates to build the best solution together

🎯
Decision mode

The team converges quickly toward a clear recommendation

4
📋 You describe your problem or question

Type in whatever you're trying to figure out — a product idea, a technical design, a content strategy, or anything else.

5
💬 Your AI team jumps into the conversation

Each specialist chimes in with their perspective, building on each other's ideas and pushing back where they disagree.

6
📝 You get a rich, multi-perspective answer

Instead of one generic response, you receive insights from every angle — risks, trade-offs, user needs, and practical next steps.

✅ You make a better decision, faster

Walk away with a well-rounded perspective that would normally take days of meetings and back-and-forth to achieve.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 25 to 25 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is ray-skills?

Ray-skills is a YAML-based role library for AI agents that enables multi-party discussions. It defines 25 distinct personas—from business analysts and system architects to content strategists and user psychologists—each with specific expertise areas, communication styles, and interaction principles. The system orchestrates these roles into structured conversations, supporting different modes like simulated debates or subagent reasoning. It integrates with AI coding tools like Claude Code, Codex, and Cursor, letting you spin up virtual teams to tackle product decisions, content planning, or technical reviews.

Why is it gaining traction?

The hook is the pre-built discussion dynamics. Instead of manually prompting each role, you get support and challenge pairs that automatically manage how personas interact—PMs push back on architects, QA questions developers, content directors challenge topic strategists. The system includes curated role lineups (like "delivery-trio" or "content-full-pipeline") so you can drop in a complete workflow instantly. For teams drowning in async back-and-forth, this turns a multi-hour Slack thread into a single AI-powered session.

Who should use this?

Product managers running feature reviews will appreciate the "alignment-trio" lineup. Content teams can leverage the "content-full-pipeline" for topic ideation through publication. Solo founders needing structured thinking without the meeting overhead will find the "fast-build-trio" useful. It's less compelling for simple tasks where a single well-crafted prompt would suffice.

Verdict

This is a promising concept with a credibility score of 0.9% and only 25 stars—early-stage and unproven at scale. The YAML configuration is clean and extensible, but documentation is minimal and the binary README limits discoverability. Worth watching if multi-agent orchestration fits your workflow, but adopt cautiously until the project gains community traction.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.