4everland/4everland-hosting-mcp

Implements a Model Context Protocol (MCP) server that exposes 4EVERLAND Hosting APIs as standardized tool calls, enabling LLM agents and AI code generators to directly invoke deployment operations without custom HTTP client code. The MCP abstraction layer translates tool schemas into backend API calls, supporting multiple decentralized storage networks (Greenfield, IPFS, Arweave) through a unified interface that abstracts network-specific implementation details.

Accepts AI-generated code artifacts (from code generation models or agents) and automatically routes them to the optimal decentralized storage backend based on file size, cost, and latency requirements. The system handles file staging, network-specific transaction preparation (gas estimation for Greenfield, IPFS pinning configuration, Arweave bundling), and returns a unified deployment result with gateway URLs and content identifiers across all backends.

Dynamically generates MCP-compliant tool schemas from 4EVERLAND Hosting API specifications and registers them with the MCP server, enabling LLM clients to discover and invoke deployment operations through standard tool-calling interfaces. The schema generation handles parameter validation, type mapping, and error response formatting to ensure LLM-safe invocation patterns.

Provides a unified abstraction layer that translates deployment requests into network-specific operations for Greenfield (BNB Chain storage), IPFS (content-addressed peer-to-peer), and Arweave (permanent storage), handling protocol differences like transaction signing, fee estimation, and content addressing. The abstraction normalizes responses across networks into a common deployment result format with network-agnostic URLs and metadata.

Tracks deployment status across Greenfield, IPFS, and Arweave networks, providing unified queries for deployment state (pending, confirmed, failed) and enabling content retrieval through network-appropriate gateways. The system maintains a deployment ledger that maps deployment IDs to network-specific identifiers and provides normalized status responses regardless of underlying network confirmation semantics.

Manages authentication credentials for 4EVERLAND Hosting and multiple decentralized storage networks (Greenfield, IPFS, Arweave), supporting multiple credential types (API keys, private keys, wallet addresses) and providing secure credential injection into deployment requests. The system handles credential rotation, expiration tracking, and network-specific authentication flows without exposing secrets to LLM clients.

Estimates deployment costs across Greenfield, IPFS, and Arweave based on file size, storage duration, and network fees, providing cost breakdowns and recommendations for backend selection. The system queries real-time or cached fee data from each network and applies heuristics to recommend the most cost-effective backend for given deployment parameters.

Manages deployment configurations and manifests that specify storage backend preferences, access controls, TTL, and other deployment parameters. The system validates configurations against schema constraints, applies defaults, and provides configuration versioning to track changes across deployments.

Implements comprehensive error handling for deployment failures across Greenfield, IPFS, and Arweave, with network-specific error classification and recovery strategies. The system provides detailed error messages, suggests remediation steps, and supports configurable retry policies with exponential backoff.