Taiwan's Open Database opens government data to AI agents, free through 2026
Taiwan's Open Database (歐噴資料庫), built by 歐噴有限公司, launched on October 1 and packages more than 3,400 Taiwan government open datasets scattered across agency websites into a structured database that AI agents can query directly, covering politics, land, companies, finance and socio-economic data. The team says each dataset records its source and field definitions and carries an "instruction manual for AI" plus human-facing dataset notes, with more than 3,000 crawlers checking sources daily for updates. Users connect through agentic tools such as Claude Code, Codex and Gemini CLI, authorize a 24-hour API token and ask questions in conversation; registration is open and free through the end of 2026, with 10 minutes of daily queries via a Google account and unlimited access after phone verification.
According to the report, the aim is to make data lookup, cross-checking and analysis — work that previously required expertise and significant time — easier through AI agents.


歐噴有限公司所開發的「歐噴資料庫」已在月初上線,將超過 3,400 份台灣政府公開資料,從散落在各機關網站、格式不一的狀態,整理成 AI Agent 能直接使用的資料庫。
歐噴資料庫將散落在數十個政府機關網站、格式不一的資料,整理成一致、可查詢的結構,並替每份資料載明來源、欄位定義,附上「給 AI 的說明書」,讓 AI Agent 知道怎麼查,也曉得什麼問題不該硬答,資料範圍涵蓋政治、土地、企業、金融、社會經濟五大領域,例如有台灣的公司資料、標案資料、社會經濟資料、立法院議案資料等等