Wispr Flow: Dictation That Works EverywhereStop typing. Start speaking. 4x faster.
Buildermark
Measure how much of your code is AI-generated. Open source.
76 followers
Measure how much of your code is AI-generated. Open source.
76 followers
Measure how much of your code is AI-generated. Buildermark matches your coding agent diffs with commits to calculate how much is by agents. Open source. Runs natively on macOS, Windows, and Linux.
Multitui
Developers and companies are claiming to have X% of their code written by AI, but most are just guessing and not actually measuring. Buildermark matches coding agent diffs with commits to calculate how much is by agents.
It also generates a convenient archive of all conversations from all of your agents, like Claude Code, Codex, Gemini, and Cursor.
Online Demo: Browse all 364 agent conversations that wrote 94% of Buildermark's code
https://demo.buildermark.dev/projects/u020uhEFtuWwPei6z6nbN
What counts as AI-generated though? Pure agent output is easy to measure. A dev who wrote the spec, accepted 90% of agent suggestions, and debugged the rest - how does your tool categorize that?
Multitui
@mykola_kondratiuk It matches all diffs in the conversation generated by the agent against committed code, so that would count 90% by the agent. Presumably, a modern workflow where the developer writes a thorough spec and the agent writes 90% of the code is still going to be more efficient and faster than traditional coding.
It’s not meant as a judgement good or bad, but more of a metric to track and compare across developers on the team, so devs can learn from each other.
Combined with ratings, you can also track which models work best in each project/language/style. I think that this will be more valuable with a future Team Server that aggregates the data for the whole team of developers.
Diff-based attribution is the right approach - deterministic and auditable. 90% lines up with what I see in most AI-assisted coding workflows too. Human ends up directing more than writing.
Hey! Looks interesting! How much of Buildermark's code is AI-generated? :)
Multitui
@antoninkus 94%! You can browse all 364 coding agent conversations that created it here: https://demo.buildermark.dev/projects/u020uhEFtuWwPei6z6nbN