solagent-forge

Expose Solana RPC tools and Anchor scaffolding through an MCP server.

Updated Feb 8, 2026
One-click install
npx skills add https://github.com/riotCode/agent-solana-project --skill solagent-forge
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: solagent-forge
Source: https://github.com/riotCode/agent-solana-project/tree/main
Command: npx skills add https://github.com/riotCode/agent-solana-project --skill solagent-forge

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

AI agents struggle to securely and deterministically access Solana blockchain primitives and scaffolding capabilities. SolAgent Forge provides an MCP server that exposes 8 focused tools for live RPC interaction (balance, accounts, transactions, program info), deterministic crypto (PDA derivation, discriminator computation), and Anchor scaffolding, enabling autonomous workflows.

Core Features & Use Cases

  • Direct Solana primitives via an MCP server with 8 tools: solana_fund_wallet, solana_get_balance, solana_get_account_info, solana_get_program_info, solana_get_transaction, solana_compute_discriminator, solana_derive_pda, anchor_scaffold.
  • Deterministic crypto for PDA derivation and instruction discriminators to reduce agent hallucinations and improve reproducibility.
  • HTTP REST wrapper for testing and MCP client integration (health, tools, and tool invocation).
  • Production-grade, deterministic tests (66 passing tests) and offline RPC-enabled workflows for reliable agent behavior.

Quick Start

Use the MCP server via npx and optionally run the HTTP wrapper for REST-style testing and tool invocation.

Frequently Asked Questions about solagent-forge

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I provide AI agents direct access to Solana RPC for querying balances and accounts?

You can provide AI agents direct Solana RPC access by running an MCP server that exposes focused tools for querying balances, accounts, transactions, and program info. This enables autonomous agents to interact with live blockchain data deterministically.

How do I compute Anchor instruction discriminators and derive PDAs without CLI dependencies?

You can compute Anchor discriminators and derive PDAs without CLI dependencies by using pure RPC-based deterministic crypto tools provided by the MCP server. This minimizes external tooling requirements and maximizes reproducibility for your automated workflows.

Can I scaffold Anchor programs for devnet and testnet using an MCP server?

Yes, you can scaffold Anchor programs for devnet and testnet workflows using the MCP server's dedicated scaffolding tool. It supports autonomous development, debugging, and auditing processes directly through agent-invoked tool calls.

Does the Solana MCP server support offline deterministic workflows for reliable agent behavior?

Yes, the Solana MCP server supports offline deterministic workflows by enforcing input validation and using pure RPC-based tooling. It includes 66 passing deterministic tests to ensure reliable and reproducible agent behavior.

What is the best way to test Solana MCP tools via HTTP REST instead of a native client?

The best way to test Solana MCP tools via HTTP REST is to use the provided HTTP REST wrapper, which exposes endpoints for health checks, listing available tools, and invoking specific tool actions for integration testing.

Why do AI agents hallucinate when deriving PDAs and how can deterministic tools prevent this?

AI agents hallucinate PDA derivation due to non-deterministic calculations, which deterministic crypto tools prevent by enforcing strict input validation and pure RPC-based computation. This ensures reproducible cryptographic results for autonomous workflows.