TECH Signal 457
Non-English agent skills share rose to 16.3% in Q2 2026, up from 13.0% in Q1
Illustration only Photo by Mika Baumeister on Unsplash
The proportion of AI agent skill files written in languages other than English grew from 13.0% to 16.3% between the first and second quarters of 2026, with Chinese accounting for the largest share.
Agent skills are plain-text SKILL.md files that the runtime multilingual model reads, so the language they are authored in can affect how precisely the agent follows instructions. A rapid rise in non-English skills signals broader global participation and may require engineers to consider localization, testing, and model language coverage. Compared with the decade-long shift in GitHub documentation, the three-point jump in three months suggests AI development is spreading beyond traditional hubs.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
Non-English skill share increased three points to 16.3% across 255,068 newly written skills in a single quarter.
Chinese dominates the non-English portion, representing 6.2% of distinct skills, while other languages each stay below 2%.
Reported English shares vary widely (65% to 99.7%) across datasets due to differing sampling methods and language-identification tools.
THE CLUSTER