Skip to content

sun-protocol/sunagent-skills

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 

Repository files navigation

SunAgent Skills

Welcome to the SunAgent Skills repository. This project serves as a centralized collection of specialized capability modules ("Skills") for AI Agents, designed to be platform-agnostic.

🧠 What is a Skill?

A Skill is a structured package of instructions, workflow rules, and resources that extends an AI agent's capabilities. It transforms a general-purpose LLM into a domain expert by providing a "Thinking Framework" and standardized procedures.

These skills are essentially system prompt extensions that can be used in:

  • Agentic IDEs: Such as Cursor, Windsurf, or custom internal tools.
  • Rules Files: By importing content into .cursorrules or similar configuration files.
  • Context Injection: By referencing the SKILL.md directly in your chat context.

Each skill typically contains:

  • SKILL.md: The core definition file containing the prompt engineering, workflow steps, and operational constraints.
  • resources/: Templates, checklists, and reference documents used by the skill.
  • scripts/: (Optional) Python or Bash scripts for automation tasks.

📂 Available Skills

Description: A rigorous, adversarial audit process specialized for DeFi protocols (DEX, Lending, Vaults, etc.).

  • Version: 2.0.0
  • Key Features:
    • 3-Phase Workflow: Automated Scan -> Manual Deep Dive -> Defense/Verdict.
    • DeFi-Specific Checks: Tailored logic for AMMs, Flash Loans, and Token Standards.
    • Test Fixture Generation: Automatically creates Foundry test bases for PoC development.
    • Invariant Verification: Uses mathematical invariants to validate or refute findings.

Description: A structured engineering workflow skill for React + TypeScript frontend development, enabling AI agents and developers to execute tasks with standardized planning, development governance, self-testing, and risk assessment.

  • Version: 1.0.0
  • Key Features:
    • 3-Phase Engineering Workflow: Planning → Development → Self-Test to ensure structured frontend task execution.
    • React + TypeScript Governance: Enforces engineering best practices including component architecture, hooks rules, state management discipline, and API/UI separation.
    • Scenario-Based Self Testing: Uses a scenario matrix to simulate QA flows, covering edge cases, negative cases, and state transitions.
    • Structured Risk Assessment: Integrates a PR Gate review process with standardized risk levels (Low / Medium / High).
    • Engineering Documentation Templates: Includes technical planning, self-test report, and PR Gate review templates for traceable engineering processes.

🚀 How to Use

For Agentic IDEs (Antigravity, Cursor etc.)

  1. Import as Rule: Copy the content of the SKILL.md file (or key sections) into your project's .cursorrules file or simply @mention the file in your chat.
  2. Prompting: Ask the agent to "Follow the workflow defined in SKILL.md to audit this project."

For Custom Agents

  1. Context Loading: Load the SKILL.md content into the agent's system prompt or context window when the relevant task is requested.
  2. Execution: Ensure the agent has access to the toolset required by the skill (e.g., file reading, terminal access).

🛠️ Contributing

To add a new skill to this repository:

  1. Create a new directory: mkdir my-new-skill.
  2. Add a SKILL.md file with the required YAML frontmatter:
    ---
    name: My New Skill
    description: A brief description of what this skill does.
    version: "1.0.0"
    ---
  3. Define your workflow instructions in standard Markdown.
  4. (Optional) Add a resources/ folder for any templates your skill needs.

About

sunagent skills

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors