CommerceAgentBench hits 1K stars: 107 e-commerce agent tasks distilled from 1.6M real conversations
VibeMarketer_ · x · 2026-09-02
CommerceAgentBench is a benchmark for commerce agents that has passed 1,000 GitHub stars and is becoming the default answer to whether an agent is ready for e-commerce work.
- Its 107 tasks span procurement, product listings, operations, fulfillment, and after-sales, mostly forcing cross-system execution: reading a messy inbox or quotes, then finishing the job in a browser, calendar, or supplier platform.
- Tasks come from real usage: Accio runs an AI agent for 10 million SMEs, and these 107 were cut from 1.6 million real conversations and 200K execution traces, backed by Alibaba's 27 years in e-commerce.
- The benchmark assumes humans keep making decisions, testing only whether an agent can handle execution underneath them.
More from coding & agent
- LLM one-shots a Minecraft 3x3 piston door by building its own Linux test infra — Vjeux · 2026-10-18
- Dev builds Mumbai-inspired 3D manga background generator from OpenStreetMap data — CurieuxExplorer · 2026-10-18
- Solo founder builds a money-printing data service on Claude's Pi Durable harness — MicahBerkley · 2026-10-18
- Dev builds Arch-based OS with custom desktop and browser using Claude Code — Snoo-10493 · 2026-10-18
- Five SaaS agent workflows fail the same way — and it's never model intelligence — usehive · 2026-10-18
- Claude skill generates hand-drawn interactive diagrams that teach kids how things work — CatAstro_Piyush · 2026-10-18