The Solana docs, rewritten for Claude Code's context window

What it does

This repository is a Claude Code skill — a structured knowledge base — that teaches the AI assistant modern Solana development practices as of January 2026. It covers the full stack, from React UI hooks and transaction building to Anchor program development, testing with LiteSVM, and security audit patterns. The repository itself is a collection of markdown reference files consumed by Claude Code’s progressive disclosure system.

The interesting bit

The skill’s real value hides in the unglamorous corners: a compatibility matrix mapping Anchor versions to Solana CLI and GLIBC releases, plus error-message-to-solution mappings for common toolchain failures. It also encodes opinionated defaults — like preferring @solana/kit over legacy web3.js and Pinocchio for compute-unit efficiency — so the AI has current best practices rather than stale stack overflow answers.

Key highlights

• Encodes a full opinionated Solana stack: framework-kit for UI, @solana/kit for clients, Anchor (or Pinocchio) for programs, and LiteSVM/Mollusk for unit tests.
• Includes specialized references for security vulnerabilities, confidential transfers, and Codama-first IDL generation.
• Maintains a version compatibility matrix and common-error troubleshooting guide to resolve toolchain mismatches.
• Uses progressive disclosure: Claude reads the main SKILL.md first, then pulls in detailed references only when relevant.

Caveats

• This is strictly a Claude Code skill; it does not provide a standalone CLI, library, or IDE extension for other editors.
• The knowledge is explicitly pinned to January 2026 best practices, so it will need refreshing as the ecosystem evolves.

Verdict

Useful if you are a Claude Code user who wants the agent to understand Solana toolchain quirks without hand-holding. If you do not use Claude Code, this skill format will not integrate with your workflow.