AGENTS.md outperforms skills in our agent evals

T1Covered by 1 source · first reported by Vercel (blog + changelog) at 27 Jan, 13:00 UTC

skills We expectedto be the solution for teaching coding agents framework-specific knowledge. After building evals focused on Next.js 16 APIs, we found something unexpected. AGENTS.md A compressed 8KB docs index embedded directly inachieved a 100% pass rate, while skills maxed out at 79% even with…

dev

Sources

• T1AGENTS.md outperforms skills in our agent evalsVercel (blog + changelog)27 Jan, 13:00