BenchDB documentation¶

BenchDB is a continuous benchmarking system for teams that need performance data to be easy to publish, inspect, compare, and act on. This reference is organized around what you are trying to do, whether you are a person using the dashboard, an automation author publishing results, or an operator running the service.

Benchmark projects send BenchDB structured result JSON. BenchDB stores those results, groups comparable measurements into history series, and helps people answer the questions that matter during development:

• what just ran,
• whether the run published the expected measurements,
• which benchmark series changed,
• whether a change is statistically meaningful,
• where to inspect the raw result, history, commit, and machine context.

The main surfaces are:

• a dashboard for browsing runs, results, series, trends, comparisons, and CI reports,
• the benchdb CLI for submitting results and generating CI diagnostics,
• OpenAPI and generated Go and TypeScript clients for automation,
• server-minted reporter tokens for CI and scripts,
• server-side alert rules for scheduled regression monitoring.

Choose your path¶

The short Guide to BenchDB follows one benchmark result through the whole system. The product page is the quickest overview for someone deciding whether BenchDB fits their team.

Supported surfaces¶

Machine-readable documentation¶

Every rendered reference page has a raw Markdown twin. Replace the trailing slash in a docs URL with .md, or start at /llms.txt for a complete index. The agent and automation guide gives coding agents the shortest reliable path through the product contracts.