Buildermark

Buildermark provides a measurement layer for coding agents, enabling users to determine the percentage of their codebase that is generated by AI coding agents. By matching coding agent logs with git commits, the tool attributes every line of code to its source without requiring workflow changes, agent hooks, or manual tagging. Buildermark is designed for developers and teams interested in understanding and tracking the influence of AI agents on their code, offering detailed agent attribution for each commit.

Key features include automatic import of chat history from supported agents such as Claude Code, Codex, Gemini, and Cursor, as well as the ability to add shared folder paths for integration with virtual machines and containers. Git history is also imported automatically, and a formatting-agnostic matcher aligns conversation diffs with commit diffs, ensuring robustness against auto-formatters and other code changes. Users can rate agent conversations manually or allow agents to log their own critiques using a specific skill. Buildermark also provides benchmarking capabilities, allowing comparison of different agents’ performance on the user’s actual codebase. Native notifications display agent attribution percentages immediately after each commit, eliminating the need for a separate dashboard.

The platform is delivered as a local-first application, running entirely on the user’s machine with no data leaving the device. It operates via a local Go server on localhost, serving a web-based user interface. Desktop applications are available for macOS (15+), Windows (10+), and Linux, with CLI support and browser extensions for Chrome, Firefox, and Safari to facilitate importing from cloud-based agents. Buildermark does not require accounts or cloud sync, and enforces zero analytics and telemetry, with the only outbound request being for update checks. The tool is MIT licensed and described as audit-friendly.

A team server feature is announced as coming soon, which will allow organizations to aggregate and compare Buildermark metrics across projects on self-hosted infrastructure. This positions Buildermark within the class of tools focused on AI code attribution and observability for both individuals and teams.