Spent 48 hours building a Model Context Protocol server for ISO 10012:2026 measurement uncertainty.
10 tools. JCGM 100:2008 plus 101:2008 compliant. Live on MCPize.
Built for ISO/IEC 17025 calibration labs that spend hours in Excel on the same uncertainty math every month. The buyer is a QA manager whose audit is in six months, not an AI engineer.
What it does
- Type A statistical uncertainty (Bessel-corrected, ν=n-1)
- Type B rectangular / triangular / k-expanded distributions
- Combined standard uncertainty u_c via root-sum-of-squares
- Welch-Satterthwaite effective degrees of freedom
- Expanded uncertainty U at target confidence level
- Monte Carlo propagation per JCGM 101:2008
- 6 pre-built uncertainty-budget templates (CMM 10mm, CD-SEM 45nm, DMM 10V, OCD 50nm, thermocouple, balance)
Try it
- Repo: https://github.com/kyb8801/measurement-uncertainty-mcp
- Live MCP: https://measurement-uncertainty.mcpize.run
- Free tier: 50 calls/month, no credit card
- Full build log (Medium): https://medium.com/@kyb8801/i-did-not-write-new-code-this-week-i-rewrote-two-paragraphs-and-ten-example-queries-d5e7aa18ecfe
Building in public. Weekly progress log: yb-ai-hustle.beehiiv.com (free).