Engineers and developers often struggle with AI assistants hallucinating material properties or providing numbers stripped of critical context. To address this, Space Mushroom has launched a free, read-only MCP (Model Context Protocol) server that connects AI tools directly to an open engineering materials database. The core value proposition is traceability: every data point returned includes the unit, test conditions, method, and a direct pointer to the source material.

Traceability Over Convenience

The server exposes eleven read-only tools designed to eliminate ambiguity in materials selection. Key functions include find_materials, which evaluates candidates against specific requirements and explicitly marks missing data as "unknown" rather than assuming failure. Another critical tool, prove_claim, returns the exact context of a value, including how it was converted from the source and where it sits within the original document. This approach ensures that when an AI cites the yield strength of 316L stainless steel, it can also show the specific ASM data sheet or standard that justified that number.

Zero-Friction Integration

Setup is intentionally low-friction for builders. The endpoint is hosted at https://spacemushroom.com/mcp using Streamable HTTP, requiring no API keys or sign-ins. For Claude Code users, integration is a single command: claude mcp add --transport http space-mushroom https://spacemushroom.com/mcp. The server is also compatible with Cursor, VS Code, and other MCP clients via a standard JSON configuration block. Additionally, it is listed in the official MCP registry as com.spacemushroom/materials, making discovery straightforward for developers browsing available servers.

Honest Data Handling

The project makes deliberate choices about data integrity that differ from typical commercial databases. Missing values are strictly labeled as "not covered yet" rather than being interpreted as safe defaults. Different test conditions are listed separately instead of being averaged out, preserving the nuance of experimental data. The database content is licensed under CC BY 4.0, and every answer includes the source for citation. A notable feature is the ai_checked label, which indicates that an AI has re-read a value against the source, but a human has not yet verified it, offering transparency about the verification level of each data point.

Key Takeaways

  • Free, open-access MCP server for engineering materials data with no authentication required.
  • Every value includes source, unit, test conditions, and method for full traceability.
  • Eleven read-only tools including find_materials, compare, and prove_claim.
  • Compatible with Claude Code, Cursor, VS Code, and listed in the official MCP registry.
  • Data licensed under CC BY 4.0 with explicit handling of missing versus failed constraints.

The Bottom Line

This is a pragmatic solution for a real pain point in engineering AI workflows. By prioritizing source transparency and refusing to average away critical test conditions, Space Mushroom offers a reliable data layer that builders can trust without needing to verify every number manually.