Blog post: what a pre-built schema saves (tokens, money, CO2)

[a6b8]

Add a public, citable blog post explaining the token/cost/CO2 savings of calling a pre-built FlowMCP schema vs. having an agent explore an API from scratch every time.

• Worked example: the 47 Berlin open-data sources from the Data Configurator (36 distinct schemas; OParl fan-out = 1 schema → 11 districts).
• All figures from a deterministic script that reads the Memo-143 measurement data + the live configurator payload (no hand-computed numbers); modelled values marked, ecological figures as ranges.
• Frames progressive disclosure / context engineering as a rationalized implementation of a known principle (Anthropic cited openly), not a novelty claim.

File: src/content/blog/2026-06-what-a-prebuilt-schema-saves.md (lang: en). Built locally with astro (177 pages, green).