Buildermark

Buildermark provides a measurement layer that links coding‑agent logs with Git commits to attribute the proportion of each change that originates from AI assistants. It automatically imports conversation histories from a range of agents such as Claude, Codex, Gemini, and Cursor, as well as the repository’s commit data, then matches diffs in a formatting‑agnostic way to determine which lines were generated by the agent. The attribution appears as a native notification at commit time, allowing developers to see AI contribution without altering their workflow or adding hooks.

The tool is designed for individual developers and teams that want visibility into how much of their codebase is produced by AI, and it supports any agent that can export chat logs. It also offers manual conversation rating and the ability to benchmark different agents against the same codebase, giving comparative insights into their performance.

Buildermark runs locally on macOS, Windows, and Linux, with a Go server serving a web UI on localhost and optional desktop or browser extensions. It stores all data on the user’s machine, includes no telemetry, and is released under the MIT license. A self‑hosted Team Server is planned for aggregating metrics across an organization.