Ask an LLM "is Node 18 still supported?" and it will answer from training data that's months or years stale. For lifecycle questions, stale means wrong — dates shift, ESU programs appear, versions die between training runs.
So we built an MCP server for endoflife.ai, now listed on Glama. Four tools:
- check_eol — is product X version Y still supported? Exact dates, days remaining or days past EOL
- get_risk_score — EOL Risk Score™ (0–100) with factor breakdown
- scan_stack — score a whole stack in one call
- list_products — resolve names across 460+ tracked products
It's a remote server — nothing to install or run:
claude mcp add --transport http endoflife https://mcp.endoflife.ai
Works without an API key. Dates come from the endoflife.date community dataset (MIT) plus our own verification layer — we cross-check against vendor pages and have had corrections merged upstream.
Try asking your assistant to scan your stack. The answer it gives you will finally have a source.
Top comments (0)