solana-dev
A unified Solana development skill hub featuring multi-agent orchestration, progressive skill loading, and deep integrations for Anchor, Token-2022, DeFi protocols, and security auditing.
Introduction
This skill acts as a comprehensive, agent-optimized configuration hub for full-stack Solana development. It is designed to minimize token usage through progressive disclosure—dynamically loading specialized sub-skills only when specific tasks require them. The agent orchestrates a ecosystem of tools, ranging from core program development (Anchor, Pinocchio) to DeFi protocol integrations (Jupiter, Drift, Raydium) and rigorous security workflows (Trail of Bits, Safe Solana Builder). It is intended for professional developers, smart contract engineers, and security auditors who need to manage complex, multi-step Solana projects efficiently.
-
Multi-Agent Workflow: Leverages 15+ specialized agents for architecture, testing, deployment, and research to automate standard development tasks.
-
Progressive Context Management: Dynamically fetches context from submodules only when necessary, keeping the workspace efficient.
-
Comprehensive Solana Tooling: Native support for Anchor v0.32+, SPL Token-2022 extensions, Formal Verification via QEDGen, and Cloudflare Workers deployment.
-
DeFi Ecosystem Integration: Direct access to specialized skills for major protocols like Kamino, Orca, Meteora, and Helius RPC.
-
Security-First Architecture: Enforces 70+ audit-derived security rules and vulnerability scanning workflows using external plugins.
-
Ideal for monorepo structures, use the primary solana-dev hub as the single entry point for all development activities.
-
Input for the agent can be specific task instructions like 'initialize a Token-2022 transfer hook' or 'audit program for reentrancy risks.'
-
The output includes synthesized code, security-hardened patterns, build instructions, and verification steps using tools like the Anchor CLI or solana-validator.
-
Ensure the repository is cloned with recursive submodules to guarantee all external skills are locally available for the agent to reference.
-
Leverage the built-in MCP server setup for real-time chain data, browser automation, and persistent memory storage.
-
Always refer to the specific subdirectory files for language-specific or protocol-specific implementation details when working on specialized components.
Repository Stats
- Stars
- 59
- Forks
- 15
- Open Issues
- 0
- Language
- Shell
- Default Branch
- main
- Sync Status
- Idle
- Last Synced
- May 3, 2026, 09:39 PM