Resonance: 65 Cross-Tool Skills to Level Up Your AI Agents

Where this fits

You're mid-sprint, juggling Claude Code for a refactor, Cursor for a UI spike, and Codex for a quick script. Each tool has its own prompt style, and you keep re-teaching them your conventions. Resonance solves this by giving you a single, portable skill library that works across all four major coding agents. Instead of writing fresh prompts for every task, you drop in a skill and get consistent, operator-grade behavior—whether you're planning a feature, writing docs, or debugging a production incident.

Why agents benefit

• Cross-tool consistency: The same skill works in Claude Code, Cursor, Codex, and opencode, so your agent's behavior doesn't change when you switch tools.
• Domain depth: 65 skills cover strategy, design, engineering, marketing, sales, and ops—not just code generation, so agents can handle full project workflows.
• Slash-command speed: 34 commands let you invoke complex workflows (e.g., /code-review, /roadmap) with a single keystroke, cutting prompt engineering time.
• Structured outputs: Skills enforce step-by-step reasoning and output formats, reducing hallucination and improving reliability on multi-step tasks.
• Maintainable updates: The library is versioned (v2.4.90) and CI-tested, so you can track changes and avoid stale prompts.

Practical scenarios

Planning a new feature end-to-end

Instead of manually prompting your agent for requirements, architecture, and test plans, use the /feature-plan command. Resonance walks the agent through discovery, edge-case analysis, and implementation steps—producing a coherent plan you can review and execute.

Standardizing code reviews across your team

Your team uses different tools. With Resonance, everyone runs /code-review and gets the same checklist: security, performance, readability, and test coverage. This removes the “my agent missed that” inconsistency and makes review feedback uniform.

Debugging a production incident under pressure

When an alert fires, you need fast, systematic triage. The /incident-response skill guides your agent to collect logs, hypothesize root causes, and propose rollback steps—without you having to craft a detailed prompt while stressed.

Add it to your agent workflow

Install Resonance with a single command:

npx skills add manusco/resonance

Then, in any supported agent (e.g., Claude Code), invoke a skill:

claude /resonance:feature-plan "Add payment webhook retry logic"

The skill loads automatically, and your agent follows the structured process. No extra configuration needed.

Compare before adopting

Before you commit, compare Resonance with alternatives like Anthropic's official skills or tool-specific prompt libraries. Check: quality signals—are the skills actively maintained and tested? Maintenance freshness—look at the release history and CI status. Alternatives—does a tool-specific solution (e.g., Cursor's built-in rules) cover your needs? Workflow fit—does the skill match how your team actually works? Resonance's breadth is its strength, but if you only use one tool, a narrower library might be simpler.

Why it is worth tracking

Resonance shows solid momentum: 36 stars, a recent v2.4.90 release, and an active CI pipeline. The fact that it's cross-tool and covers non-engineering domains makes it a rare find. If you're building agents that span multiple tools or need consistent behavior across your team, this is worth evaluating now. For a deeper look at its skill list and community feedback, see the OpenAgentSkill entry. As the library grows, it could become a default layer for agent workflows—so tracking its evolution is smart even if you don't adopt it today.