6.9k stars · MIT · @evidence-dev/evidence@40.1.8 (2026-02-06) · Track this in Scout
Business intelligence as Markdown files with SQL in them, building to a static site — a dashboard Claude Code can write.
▶Repo detailsthe review · specs · pros & cons
What it is
A business-intelligence tool where each page is a Markdown file containing SQL queries and chart components, and the build produces a fast static website. It connects to PostgreSQL, SQLite, CSV files and more.
What it is good for. By the end of this edition the numbers live in four places: Umami's tables, Listmonk's tables, the research SQLite file from 05, and the application's own database. The question worth answering — did the video that went out on Tuesday produce signups on Wednesday — spans all of them, and no single tool's dashboard will ever show it. Evidence is the one shape of BI tool that suits an agent-written codebase: the whole dashboard is text files in a git repository, which means Claude Code can write and revise it the same way it writes the application, and the output is a static site that can sit on the same nginx that serves everything else.
- Text files in git. No dragging widgets, no clicking through a builder, no state trapped in someone's UI.
- Static output — nothing to keep running, nothing to secure beyond a password on a directory.
- Talks to PostgreSQL and SQLite natively, which is exactly the pair this edition ends up with.
- The Node/npm build toolchain is a second ecosystem to keep working next to a Python one, and version drift there is a familiar tax.
- It queries at build time, so a "live" dashboard means a scheduled rebuild — fine daily, wrong for anything you want to watch in real time.
- Honestly premature until 04 and Umami are producing data. Read this entry, do not install it this week.
